1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs -fexternal-templates @gol
178 -falt-external-templates @gol
179 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
180 -fno-implicit-templates @gol
181 -fno-implicit-inline-templates @gol
182 -fno-implement-inlines -fms-extensions @gol
183 -fno-nonansi-builtins -fno-operator-names @gol
184 -fno-optional-diags -fpermissive @gol
185 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
186 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
187 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C Language Options
195 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime -gen-decls @gol
198 -Wno-protocol -Wselector -Wundeclared-selector}
200 @item Language Independent Options
201 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 @gccoptlist{-fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
207 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
208 -w -Wextra -Wall -Waggregate-return @gol
209 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
210 -Wconversion -Wno-deprecated-declarations @gol
211 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Wimport -Winline -Winvalid-pch -Wno-endif-labels @gol
218 -Wno-invalid-offsetof @gol
219 -Wlarger-than-@var{len} -Wlong-long @gol
220 -Wmain -Wmissing-braces @gol
221 -Wmissing-format-attribute -Wmissing-noreturn @gol
222 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
223 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
224 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
225 -Wreturn-type -Wsequence-point -Wshadow @gol
226 -Wsign-compare -Wstrict-aliasing @gol
227 -Wswitch -Wswitch-default -Wswitch-enum @gol
228 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
229 -Wunknown-pragmas -Wunreachable-code @gol
230 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
231 -Wunused-value -Wunused-variable -Wwrite-strings}
233 @item C-only Warning Options
234 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
235 -Wmissing-prototypes -Wnested-externs @gol
236 -Wstrict-prototypes -Wtraditional}
238 @item Debugging Options
239 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
240 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
241 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
242 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
244 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
245 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
246 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
247 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
248 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
249 -ftest-coverage -ftime-report @gol
250 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
251 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
252 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
253 -print-multi-directory -print-multi-lib @gol
254 -print-prog-name=@var{program} -print-search-dirs -Q @gol
257 @item Optimization Options
258 @xref{Optimize Options,,Options that Control Optimization}.
259 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
260 -falign-labels=@var{n} -falign-loops=@var{n} @gol
261 -fbranch-probabilities -fprofile-values -fbranch-target-load-optimize @gol
262 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
263 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
264 -fdelayed-branch -fdelete-null-pointer-checks @gol
265 -fexpensive-optimizations -ffast-math -ffloat-store @gol
266 -fforce-addr -fforce-mem -ffunction-sections @gol
267 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
268 -fif-conversion -fif-conversion2 @gol
269 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
270 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
271 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
272 -fno-default-inline -fno-defer-pop @gol
273 -fno-function-cse -fno-guess-branch-probability @gol
274 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
275 -funsafe-math-optimizations -ffinite-math-only @gol
276 -fno-trapping-math -fno-zero-initialized-in-bss @gol
277 -fomit-frame-pointer -foptimize-register-move @gol
278 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
279 -freduce-all-givs -fregmove -frename-registers @gol
280 -freorder-blocks -freorder-functions @gol
281 -frerun-cse-after-loop -frerun-loop-opt @gol
282 -fschedule-insns -fschedule-insns2 @gol
283 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
284 -fsched-spec-load-dangerous -fsched2-use-superblocks @gol
285 -fsched2-use-traces -fsignaling-nans @gol
286 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
287 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
288 -funroll-all-loops -funroll-loops -fpeel-loops @gol
289 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
290 --param @var{name}=@var{value}
291 -O -O0 -O1 -O2 -O3 -Os}
293 @item Preprocessor Options
294 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
295 @gccoptlist{-A@var{question}=@var{answer} @gol
296 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
297 -C -dD -dI -dM -dN @gol
298 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
299 -idirafter @var{dir} @gol
300 -include @var{file} -imacros @var{file} @gol
301 -iprefix @var{file} -iwithprefix @var{dir} @gol
302 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
303 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
304 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
305 -Xpreprocessor @var{option}}
307 @item Assembler Option
308 @xref{Assembler Options,,Passing Options to the Assembler}.
309 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
312 @xref{Link Options,,Options for Linking}.
313 @gccoptlist{@var{object-file-name} -l@var{library} @gol
314 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
315 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
316 -Wl,@var{option} -Xlinker @var{option} @gol
319 @item Directory Options
320 @xref{Directory Options,,Options for Directory Search}.
321 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
324 @c I wrote this xref this way to avoid overfull hbox. -- rms
325 @xref{Target Options}.
326 @gccoptlist{-V @var{version} -b @var{machine}}
328 @item Machine Dependent Options
329 @xref{Submodel Options,,Hardware Models and Configurations}.
331 @emph{M680x0 Options}
332 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
333 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
334 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
335 -malign-int -mstrict-align}
337 @emph{M68hc1x Options}
338 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
339 -mauto-incdec -minmax -mlong-calls -mshort @gol
340 -msoft-reg-count=@var{count}}
343 @gccoptlist{-mg -mgnu -munix}
346 @gccoptlist{-mcpu=@var{cpu-type} @gol
347 -mtune=@var{cpu-type} @gol
348 -mcmodel=@var{code-model} @gol
350 -mapp-regs -mbroken-saverestore -mcypress @gol
351 -mfaster-structs -mflat @gol
352 -mfpu -mhard-float -mhard-quad-float @gol
353 -mimpure-text -mlive-g0 -mno-app-regs @gol
354 -mno-faster-structs -mno-flat -mno-fpu @gol
355 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
356 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
357 -msupersparc -munaligned-doubles -mv8}
360 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
361 -mapcs-26 -mapcs-32 @gol
362 -mapcs-stack-check -mno-apcs-stack-check @gol
363 -mapcs-float -mno-apcs-float @gol
364 -mapcs-reentrant -mno-apcs-reentrant @gol
365 -msched-prolog -mno-sched-prolog @gol
366 -mlittle-endian -mbig-endian -mwords-little-endian @gol
367 -malignment-traps -mno-alignment-traps @gol
368 -msoft-float -mhard-float -mfpe @gol
369 -mthumb-interwork -mno-thumb-interwork @gol
370 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
371 -mstructure-size-boundary=@var{n} @gol
372 -mabort-on-noreturn @gol
373 -mlong-calls -mno-long-calls @gol
374 -msingle-pic-base -mno-single-pic-base @gol
375 -mpic-register=@var{reg} @gol
376 -mnop-fun-dllimport @gol
377 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
378 -mpoke-function-name @gol
380 -mtpcs-frame -mtpcs-leaf-frame @gol
381 -mcaller-super-interworking -mcallee-super-interworking}
383 @emph{MN10200 Options}
386 @emph{MN10300 Options}
387 @gccoptlist{-mmult-bug -mno-mult-bug @gol
388 -mam33 -mno-am33 @gol
391 @emph{M32R/D Options}
392 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
393 -msdata=@var{sdata-type} -G @var{num}}
396 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
397 -mcheck-zero-division -mhandle-large-shift @gol
398 -midentify-revision -mno-check-zero-division @gol
399 -mno-ocs-debug-info -mno-ocs-frame-position @gol
400 -mno-optimize-arg-area -mno-serialize-volatile @gol
401 -mno-underscores -mocs-debug-info @gol
402 -mocs-frame-position -moptimize-arg-area @gol
403 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
404 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
405 -mversion-03.00 -mwarn-passed-structs}
407 @emph{RS/6000 and PowerPC Options}
408 @gccoptlist{-mcpu=@var{cpu-type} @gol
409 -mtune=@var{cpu-type} @gol
410 -mpower -mno-power -mpower2 -mno-power2 @gol
411 -mpowerpc -mpowerpc64 -mno-powerpc @gol
412 -maltivec -mno-altivec @gol
413 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
414 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
415 -mnew-mnemonics -mold-mnemonics @gol
416 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
417 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
418 -malign-power -malign-natural @gol
419 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
420 -mstring -mno-string -mupdate -mno-update @gol
421 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
422 -mstrict-align -mno-strict-align -mrelocatable @gol
423 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
424 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
425 -mdynamic-no-pic @gol
426 -mcall-sysv -mcall-netbsd @gol
427 -maix-struct-return -msvr4-struct-return @gol
428 -mabi=altivec -mabi=no-altivec @gol
429 -mabi=spe -mabi=no-spe @gol
430 -misel=yes -misel=no @gol
431 -mspe=yes -mspe=no @gol
432 -mfloat-gprs=yes -mfloat-gprs=no @gol
433 -mprototype -mno-prototype @gol
434 -msim -mmvme -mads -myellowknife -memb -msdata @gol
435 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
437 @emph{Darwin Options}
438 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
439 -arch_only -bind_at_load -bundle -bundle_loader @gol
440 -client_name -compatibility_version -current_version @gol
441 -dependency-file -dylib_file -dylinker_install_name @gol
442 -dynamic -dynamiclib -exported_symbols_list @gol
443 -filelist -flat_namespace -force_cpusubtype_ALL @gol
444 -force_flat_namespace -headerpad_max_install_names @gol
445 -image_base -init -install_name -keep_private_externs @gol
446 -multi_module -multiply_defined -multiply_defined_unused @gol
447 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
448 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
449 -private_bundle -read_only_relocs -sectalign @gol
450 -sectobjectsymbols -whyload -seg1addr @gol
451 -sectcreate -sectobjectsymbols -sectorder @gol
452 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
453 -segprot -segs_read_only_addr -segs_read_write_addr @gol
454 -single_module -static -sub_library -sub_umbrella @gol
455 -twolevel_namespace -umbrella -undefined @gol
456 -unexported_symbols_list -weak_reference_mismatches @gol
460 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
461 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
462 -mminimum-fp-blocks -mnohc-struct-return}
465 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
466 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
467 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
468 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
469 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
470 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
471 -mmips-as -mmips-tfile -mno-abicalls @gol
472 -mno-embedded-data -mno-uninit-const-in-rodata @gol
473 -mno-embedded-pic -mno-long-calls @gol
474 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
475 -mrnames -msoft-float @gol
476 -m4650 -msingle-float -mmad @gol
477 -EL -EB -G @var{num} -nocpp @gol
478 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
479 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
480 -mbranch-likely -mno-branch-likely}
482 @emph{i386 and x86-64 Options}
483 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
484 -mfpmath=@var{unit} @gol
485 -masm=@var{dialect} -mno-fancy-math-387 @gol
486 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
487 -mno-wide-multiply -mrtd -malign-double @gol
488 -mpreferred-stack-boundary=@var{num} @gol
489 -mmmx -msse -msse2 -mpni -m3dnow @gol
490 -mthreads -mno-align-stringops -minline-all-stringops @gol
491 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
492 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
493 -mno-red-zone -mno-tls-direct-seg-refs @gol
494 -mcmodel=@var{code-model} @gol
498 @gccoptlist{-march=@var{architecture-type} @gol
499 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
500 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
501 -mjump-in-delay -mlinker-opt -mlong-calls @gol
502 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
503 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
504 -mno-jump-in-delay -mno-long-load-store @gol
505 -mno-portable-runtime -mno-soft-float @gol
506 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
507 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
508 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
509 -nolibdld -static -threads}
511 @emph{Intel 960 Options}
512 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
513 -mcode-align -mcomplex-addr -mleaf-procedures @gol
514 -mic-compat -mic2.0-compat -mic3.0-compat @gol
515 -mintel-asm -mno-clean-linkage -mno-code-align @gol
516 -mno-complex-addr -mno-leaf-procedures @gol
517 -mno-old-align -mno-strict-align -mno-tail-call @gol
518 -mnumerics -mold-align -msoft-float -mstrict-align @gol
521 @emph{DEC Alpha Options}
522 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
523 -mieee -mieee-with-inexact -mieee-conformant @gol
524 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
525 -mtrap-precision=@var{mode} -mbuild-constants @gol
526 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
527 -mbwx -mmax -mfix -mcix @gol
528 -mfloat-vax -mfloat-ieee @gol
529 -mexplicit-relocs -msmall-data -mlarge-data @gol
530 -msmall-text -mlarge-text @gol
531 -mmemory-latency=@var{time}}
533 @emph{DEC Alpha/VMS Options}
534 @gccoptlist{-mvms-return-codes}
536 @emph{H8/300 Options}
537 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
540 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
541 -m4-nofpu -m4-single-only -m4-single -m4 @gol
542 -m5-64media -m5-64media-nofpu @gol
543 -m5-32media -m5-32media-nofpu @gol
544 -m5-compact -m5-compact-nofpu @gol
545 -mb -ml -mdalign -mrelax @gol
546 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
547 -mieee -misize -mpadstruct -mspace @gol
548 -mprefergot -musermode}
550 @emph{System V Options}
551 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
554 @gccoptlist{-EB -EL @gol
555 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
556 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
558 @emph{TMS320C3x/C4x Options}
559 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
560 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
561 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
562 -mparallel-insns -mparallel-mpy -mpreserve-float}
565 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
566 -mprolog-function -mno-prolog-function -mspace @gol
567 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
568 -mapp-regs -mno-app-regs @gol
569 -mdisable-callt -mno-disable-callt @gol
574 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
575 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
576 -mregparam -mnoregparam -msb -mnosb @gol
577 -mbitfield -mnobitfield -mhimem -mnohimem}
580 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
581 -mcall-prologues -mno-tablejump -mtiny-stack}
584 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
585 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
586 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
587 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
588 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
591 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
592 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
593 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
594 -mno-base-addresses -msingle-exit -mno-single-exit}
597 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
598 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
599 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
600 -minline-float-divide-max-throughput @gol
601 -minline-int-divide-min-latency @gol
602 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
603 -mfixed-range=@var{register-range}}
606 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
607 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
609 @emph{S/390 and zSeries Options}
610 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
611 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
612 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
613 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
616 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
617 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
618 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
619 -mstack-align -mdata-align -mconst-align @gol
620 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
621 -melf -maout -melinux -mlinux -sim -sim2}
623 @emph{PDP-11 Options}
624 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
625 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
626 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
627 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
628 -mbranch-expensive -mbranch-cheap @gol
629 -msplit -mno-split -munix-asm -mdec-asm}
631 @emph{Xstormy16 Options}
634 @emph{Xtensa Options}
635 @gccoptlist{-mbig-endian -mlittle-endian @gol
636 -mdensity -mno-density @gol
637 -mconst16 -mno-const16 @gol
639 -maddx -mno-addx @gol
640 -mmac16 -mno-mac16 @gol
641 -mmul16 -mno-mul16 @gol
642 -mmul32 -mno-mul32 @gol
644 -mminmax -mno-minmax @gol
645 -msext -mno-sext @gol
646 -mbooleans -mno-booleans @gol
647 -mhard-float -msoft-float @gol
648 -mfused-madd -mno-fused-madd @gol
649 -mtext-section-literals -mno-text-section-literals @gol
650 -mtarget-align -mno-target-align @gol
651 -mlongcalls -mno-longcalls}
654 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
655 -mhard-float -msoft-float @gol
656 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
657 -mdouble -mno-double @gol
658 -mmedia -mno-media -mmuladd -mno-muladd @gol
659 -mlibrary-pic -macc-4 -macc-8 @gol
660 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
661 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
662 -mvliw-branch -mno-vliw-branch @gol
663 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
664 -mno-nested-cond-exec -mtomcat-stats @gol
667 @item Code Generation Options
668 @xref{Code Gen Options,,Options for Code Generation Conventions}.
669 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
670 -ffixed-@var{reg} -fexceptions @gol
671 -fnon-call-exceptions -funwind-tables @gol
672 -fasynchronous-unwind-tables @gol
673 -finhibit-size-directive -finstrument-functions @gol
674 -fno-common -fno-ident -fno-gnu-linker @gol
675 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
676 -freg-struct-return -fshared-data -fshort-enums @gol
677 -fshort-double -fshort-wchar @gol
678 -fverbose-asm -fpack-struct -fstack-check @gol
679 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
680 -fargument-alias -fargument-noalias @gol
681 -fargument-noalias-global -fleading-underscore @gol
682 -ftls-model=@var{model} @gol
683 -ftrapv -fwrapv -fbounds-check}
687 * Overall Options:: Controlling the kind of output:
688 an executable, object files, assembler files,
689 or preprocessed source.
690 * C Dialect Options:: Controlling the variant of C language compiled.
691 * C++ Dialect Options:: Variations on C++.
692 * Objective-C Dialect Options:: Variations on Objective-C.
693 * Language Independent Options:: Controlling how diagnostics should be
695 * Warning Options:: How picky should the compiler be?
696 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
697 * Optimize Options:: How much optimization?
698 * Preprocessor Options:: Controlling header files and macro definitions.
699 Also, getting dependency information for Make.
700 * Assembler Options:: Passing options to the assembler.
701 * Link Options:: Specifying libraries and so on.
702 * Directory Options:: Where to find header files and libraries.
703 Where to find the compiler executable files.
704 * Spec Files:: How to pass switches to sub-processes.
705 * Target Options:: Running a cross-compiler, or an old version of GCC.
708 @node Overall Options
709 @section Options Controlling the Kind of Output
711 Compilation can involve up to four stages: preprocessing, compilation
712 proper, assembly and linking, always in that order. The first three
713 stages apply to an individual source file, and end by producing an
714 object file; linking combines all the object files (those newly
715 compiled, and those specified as input) into an executable file.
717 @cindex file name suffix
718 For any given input file, the file name suffix determines what kind of
723 C source code which must be preprocessed.
726 C source code which should not be preprocessed.
729 C++ source code which should not be preprocessed.
732 Objective-C source code. Note that you must link with the library
733 @file{libobjc.a} to make an Objective-C program work.
736 Objective-C source code which should not be preprocessed.
739 C or C++ header file to be turned into a precompiled header.
743 @itemx @var{file}.cxx
744 @itemx @var{file}.cpp
745 @itemx @var{file}.CPP
746 @itemx @var{file}.c++
748 C++ source code which must be preprocessed. Note that in @samp{.cxx},
749 the last two letters must both be literally @samp{x}. Likewise,
750 @samp{.C} refers to a literal capital C@.
754 C++ header file to be turned into a precompiled header.
757 @itemx @var{file}.for
758 @itemx @var{file}.FOR
759 Fortran source code which should not be preprocessed.
762 @itemx @var{file}.fpp
763 @itemx @var{file}.FPP
764 Fortran source code which must be preprocessed (with the traditional
768 Fortran source code which must be preprocessed with a RATFOR
769 preprocessor (not included with GCC)@.
771 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
772 Using and Porting GNU Fortran}, for more details of the handling of
775 @c FIXME: Descriptions of Java file types.
782 Ada source code file which contains a library unit declaration (a
783 declaration of a package, subprogram, or generic, or a generic
784 instantiation), or a library unit renaming declaration (a package,
785 generic, or subprogram renaming declaration). Such files are also
788 @itemx @var{file}.adb
789 Ada source code file containing a library unit body (a subprogram or
790 package body). Such files are also called @dfn{bodies}.
792 @c GCC also knows about some suffixes for languages not yet included:
801 Assembler code which must be preprocessed.
804 An object file to be fed straight into linking.
805 Any file name with no recognized suffix is treated this way.
809 You can specify the input language explicitly with the @option{-x} option:
812 @item -x @var{language}
813 Specify explicitly the @var{language} for the following input files
814 (rather than letting the compiler choose a default based on the file
815 name suffix). This option applies to all following input files until
816 the next @option{-x} option. Possible values for @var{language} are:
818 c c-header cpp-output
819 c++ c++-header c++-cpp-output
820 objective-c objective-c-header objc-cpp-output
821 assembler assembler-with-cpp
823 f77 f77-cpp-input ratfor
829 Turn off any specification of a language, so that subsequent files are
830 handled according to their file name suffixes (as they are if @option{-x}
831 has not been used at all).
833 @item -pass-exit-codes
834 @opindex pass-exit-codes
835 Normally the @command{gcc} program will exit with the code of 1 if any
836 phase of the compiler returns a non-success return code. If you specify
837 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
838 numerically highest error produced by any phase that returned an error
842 If you only want some of the stages of compilation, you can use
843 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
844 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
845 @command{gcc} is to stop. Note that some combinations (for example,
846 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
851 Compile or assemble the source files, but do not link. The linking
852 stage simply is not done. The ultimate output is in the form of an
853 object file for each source file.
855 By default, the object file name for a source file is made by replacing
856 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
858 Unrecognized input files, not requiring compilation or assembly, are
863 Stop after the stage of compilation proper; do not assemble. The output
864 is in the form of an assembler code file for each non-assembler input
867 By default, the assembler file name for a source file is made by
868 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
870 Input files that don't require compilation are ignored.
874 Stop after the preprocessing stage; do not run the compiler proper. The
875 output is in the form of preprocessed source code, which is sent to the
878 Input files which don't require preprocessing are ignored.
880 @cindex output file option
883 Place output in file @var{file}. This applies regardless to whatever
884 sort of output is being produced, whether it be an executable file,
885 an object file, an assembler file or preprocessed C code.
887 Since only one output file can be specified, it does not make sense to
888 use @option{-o} when compiling more than one input file, unless you are
889 producing an executable file as output.
891 If @option{-o} is not specified, the default is to put an executable file
892 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
893 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
894 all preprocessed C source on standard output.
898 Print (on standard error output) the commands executed to run the stages
899 of compilation. Also print the version number of the compiler driver
900 program and of the preprocessor and the compiler proper.
904 Like @option{-v} except the commands are not executed and all command
905 arguments are quoted. This is useful for shell scripts to capture the
906 driver-generated command lines.
910 Use pipes rather than temporary files for communication between the
911 various stages of compilation. This fails to work on some systems where
912 the assembler is unable to read from a pipe; but the GNU assembler has
917 Print (on the standard output) a description of the command line options
918 understood by @command{gcc}. If the @option{-v} option is also specified
919 then @option{--help} will also be passed on to the various processes
920 invoked by @command{gcc}, so that they can display the command line options
921 they accept. If the @option{-Wextra} option is also specified then command
922 line options which have no documentation associated with them will also
927 Print (on the standard output) a description of target specific command
928 line options for each tool.
932 Display the version number and copyrights of the invoked GCC.
936 @section Compiling C++ Programs
938 @cindex suffixes for C++ source
939 @cindex C++ source file suffixes
940 C++ source files conventionally use one of the suffixes @samp{.C},
941 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
942 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
943 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
944 files with these names and compiles them as C++ programs even if you
945 call the compiler the same way as for compiling C programs (usually
946 with the name @command{gcc}).
950 However, C++ programs often require class libraries as well as a
951 compiler that understands the C++ language---and under some
952 circumstances, you might want to compile programs or header files from
953 standard input, or otherwise without a suffix that flags them as C++
954 programs. You might also like to precompile a C header file with a
955 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
956 program that calls GCC with the default language set to C++, and
957 automatically specifies linking against the C++ library. On many
958 systems, @command{g++} is also installed with the name @command{c++}.
960 @cindex invoking @command{g++}
961 When you compile C++ programs, you may specify many of the same
962 command-line options that you use for compiling programs in any
963 language; or command-line options meaningful for C and related
964 languages; or options that are meaningful only for C++ programs.
965 @xref{C Dialect Options,,Options Controlling C Dialect}, for
966 explanations of options for languages related to C@.
967 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
968 explanations of options that are meaningful only for C++ programs.
970 @node C Dialect Options
971 @section Options Controlling C Dialect
972 @cindex dialect options
973 @cindex language dialect options
974 @cindex options, dialect
976 The following options control the dialect of C (or languages derived
977 from C, such as C++ and Objective-C) that the compiler accepts:
984 In C mode, support all ISO C90 programs. In C++ mode,
985 remove GNU extensions that conflict with ISO C++.
987 This turns off certain features of GCC that are incompatible with ISO
988 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
989 such as the @code{asm} and @code{typeof} keywords, and
990 predefined macros such as @code{unix} and @code{vax} that identify the
991 type of system you are using. It also enables the undesirable and
992 rarely used ISO trigraph feature. For the C compiler,
993 it disables recognition of C++ style @samp{//} comments as well as
994 the @code{inline} keyword.
996 The alternate keywords @code{__asm__}, @code{__extension__},
997 @code{__inline__} and @code{__typeof__} continue to work despite
998 @option{-ansi}. You would not want to use them in an ISO C program, of
999 course, but it is useful to put them in header files that might be included
1000 in compilations done with @option{-ansi}. Alternate predefined macros
1001 such as @code{__unix__} and @code{__vax__} are also available, with or
1002 without @option{-ansi}.
1004 The @option{-ansi} option does not cause non-ISO programs to be
1005 rejected gratuitously. For that, @option{-pedantic} is required in
1006 addition to @option{-ansi}. @xref{Warning Options}.
1008 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1009 option is used. Some header files may notice this macro and refrain
1010 from declaring certain functions or defining certain macros that the
1011 ISO standard doesn't call for; this is to avoid interfering with any
1012 programs that might use these names for other things.
1014 Functions which would normally be built in but do not have semantics
1015 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1016 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1017 built-in functions provided by GCC}, for details of the functions
1022 Determine the language standard. This option is currently only
1023 supported when compiling C or C++. A value for this option must be
1024 provided; possible values are
1029 ISO C90 (same as @option{-ansi}).
1031 @item iso9899:199409
1032 ISO C90 as modified in amendment 1.
1038 ISO C99. Note that this standard is not yet fully supported; see
1039 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1040 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1043 Default, ISO C90 plus GNU extensions (including some C99 features).
1047 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1048 this will become the default. The name @samp{gnu9x} is deprecated.
1051 The 1998 ISO C++ standard plus amendments.
1054 The same as @option{-std=c++98} plus GNU extensions. This is the
1055 default for C++ code.
1058 Even when this option is not specified, you can still use some of the
1059 features of newer standards in so far as they do not conflict with
1060 previous C standards. For example, you may use @code{__restrict__} even
1061 when @option{-std=c99} is not specified.
1063 The @option{-std} options specifying some version of ISO C have the same
1064 effects as @option{-ansi}, except that features that were not in ISO C90
1065 but are in the specified version (for example, @samp{//} comments and
1066 the @code{inline} keyword in ISO C99) are not disabled.
1068 @xref{Standards,,Language Standards Supported by GCC}, for details of
1069 these standard versions.
1071 @item -aux-info @var{filename}
1073 Output to the given filename prototyped declarations for all functions
1074 declared and/or defined in a translation unit, including those in header
1075 files. This option is silently ignored in any language other than C@.
1077 Besides declarations, the file indicates, in comments, the origin of
1078 each declaration (source file and line), whether the declaration was
1079 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1080 @samp{O} for old, respectively, in the first character after the line
1081 number and the colon), and whether it came from a declaration or a
1082 definition (@samp{C} or @samp{F}, respectively, in the following
1083 character). In the case of function definitions, a K&R-style list of
1084 arguments followed by their declarations is also provided, inside
1085 comments, after the declaration.
1089 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1090 keyword, so that code can use these words as identifiers. You can use
1091 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1092 instead. @option{-ansi} implies @option{-fno-asm}.
1094 In C++, this switch only affects the @code{typeof} keyword, since
1095 @code{asm} and @code{inline} are standard keywords. You may want to
1096 use the @option{-fno-gnu-keywords} flag instead, which has the same
1097 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1098 switch only affects the @code{asm} and @code{typeof} keywords, since
1099 @code{inline} is a standard keyword in ISO C99.
1102 @itemx -fno-builtin-@var{function}
1103 @opindex fno-builtin
1104 @cindex built-in functions
1105 Don't recognize built-in functions that do not begin with
1106 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1107 functions provided by GCC}, for details of the functions affected,
1108 including those which are not built-in functions when @option{-ansi} or
1109 @option{-std} options for strict ISO C conformance are used because they
1110 do not have an ISO standard meaning.
1112 GCC normally generates special code to handle certain built-in functions
1113 more efficiently; for instance, calls to @code{alloca} may become single
1114 instructions that adjust the stack directly, and calls to @code{memcpy}
1115 may become inline copy loops. The resulting code is often both smaller
1116 and faster, but since the function calls no longer appear as such, you
1117 cannot set a breakpoint on those calls, nor can you change the behavior
1118 of the functions by linking with a different library.
1120 With the @option{-fno-builtin-@var{function}} option
1121 only the built-in function @var{function} is
1122 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1123 function is named this is not built-in in this version of GCC, this
1124 option is ignored. There is no corresponding
1125 @option{-fbuiltin-@var{function}} option; if you wish to enable
1126 built-in functions selectively when using @option{-fno-builtin} or
1127 @option{-ffreestanding}, you may define macros such as:
1130 #define abs(n) __builtin_abs ((n))
1131 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1136 @cindex hosted environment
1138 Assert that compilation takes place in a hosted environment. This implies
1139 @option{-fbuiltin}. A hosted environment is one in which the
1140 entire standard library is available, and in which @code{main} has a return
1141 type of @code{int}. Examples are nearly everything except a kernel.
1142 This is equivalent to @option{-fno-freestanding}.
1144 @item -ffreestanding
1145 @opindex ffreestanding
1146 @cindex hosted environment
1148 Assert that compilation takes place in a freestanding environment. This
1149 implies @option{-fno-builtin}. A freestanding environment
1150 is one in which the standard library may not exist, and program startup may
1151 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1152 This is equivalent to @option{-fno-hosted}.
1154 @xref{Standards,,Language Standards Supported by GCC}, for details of
1155 freestanding and hosted environments.
1157 @item -fms-extensions
1158 @opindex fms-extensions
1159 Accept some non-standard constructs used in Microsoft header files.
1163 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1164 options for strict ISO C conformance) implies @option{-trigraphs}.
1166 @item -no-integrated-cpp
1167 @opindex no-integrated-cpp
1168 Performs a compilation in two passes: preprocessing and compiling. This
1169 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1170 @option{-B} option. The user supplied compilation step can then add in
1171 an additional preprocessing step after normal preprocessing but before
1172 compiling. The default is to use the integrated cpp (internal cpp)
1174 The semantics of this option will change if "cc1", "cc1plus", and
1175 "cc1obj" are merged.
1177 @cindex traditional C language
1178 @cindex C language, traditional
1180 @itemx -traditional-cpp
1181 @opindex traditional-cpp
1182 @opindex traditional
1183 Formerly, these options caused GCC to attempt to emulate a pre-standard
1184 C compiler. They are now only supported with the @option{-E} switch.
1185 The preprocessor continues to support a pre-standard mode. See the GNU
1186 CPP manual for details.
1188 @item -fcond-mismatch
1189 @opindex fcond-mismatch
1190 Allow conditional expressions with mismatched types in the second and
1191 third arguments. The value of such an expression is void. This option
1192 is not supported for C++.
1194 @item -funsigned-char
1195 @opindex funsigned-char
1196 Let the type @code{char} be unsigned, like @code{unsigned char}.
1198 Each kind of machine has a default for what @code{char} should
1199 be. It is either like @code{unsigned char} by default or like
1200 @code{signed char} by default.
1202 Ideally, a portable program should always use @code{signed char} or
1203 @code{unsigned char} when it depends on the signedness of an object.
1204 But many programs have been written to use plain @code{char} and
1205 expect it to be signed, or expect it to be unsigned, depending on the
1206 machines they were written for. This option, and its inverse, let you
1207 make such a program work with the opposite default.
1209 The type @code{char} is always a distinct type from each of
1210 @code{signed char} or @code{unsigned char}, even though its behavior
1211 is always just like one of those two.
1214 @opindex fsigned-char
1215 Let the type @code{char} be signed, like @code{signed char}.
1217 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1218 the negative form of @option{-funsigned-char}. Likewise, the option
1219 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1221 @item -fsigned-bitfields
1222 @itemx -funsigned-bitfields
1223 @itemx -fno-signed-bitfields
1224 @itemx -fno-unsigned-bitfields
1225 @opindex fsigned-bitfields
1226 @opindex funsigned-bitfields
1227 @opindex fno-signed-bitfields
1228 @opindex fno-unsigned-bitfields
1229 These options control whether a bit-field is signed or unsigned, when the
1230 declaration does not use either @code{signed} or @code{unsigned}. By
1231 default, such a bit-field is signed, because this is consistent: the
1232 basic integer types such as @code{int} are signed types.
1234 @item -fwritable-strings
1235 @opindex fwritable-strings
1236 Store string constants in the writable data segment and don't uniquize
1237 them. This is for compatibility with old programs which assume they can
1238 write into string constants.
1240 Writing into string constants is a very bad idea; ``constants'' should
1244 @node C++ Dialect Options
1245 @section Options Controlling C++ Dialect
1247 @cindex compiler options, C++
1248 @cindex C++ options, command line
1249 @cindex options, C++
1250 This section describes the command-line options that are only meaningful
1251 for C++ programs; but you can also use most of the GNU compiler options
1252 regardless of what language your program is in. For example, you
1253 might compile a file @code{firstClass.C} like this:
1256 g++ -g -frepo -O -c firstClass.C
1260 In this example, only @option{-frepo} is an option meant
1261 only for C++ programs; you can use the other options with any
1262 language supported by GCC@.
1264 Here is a list of options that are @emph{only} for compiling C++ programs:
1268 @item -fabi-version=@var{n}
1269 @opindex fabi-version
1270 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1271 ABI that first appeared in G++ 3.2. Version 0 will always be the
1272 version that conforms most closely to the C++ ABI specification.
1273 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1276 The default is version 1.
1278 @item -fno-access-control
1279 @opindex fno-access-control
1280 Turn off all access checking. This switch is mainly useful for working
1281 around bugs in the access control code.
1285 Check that the pointer returned by @code{operator new} is non-null
1286 before attempting to modify the storage allocated. This check is
1287 normally unnecessary because the C++ standard specifies that
1288 @code{operator new} will only return @code{0} if it is declared
1289 @samp{throw()}, in which case the compiler will always check the
1290 return value even without this option. In all other cases, when
1291 @code{operator new} has a non-empty exception specification, memory
1292 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1293 @samp{new (nothrow)}.
1295 @item -fconserve-space
1296 @opindex fconserve-space
1297 Put uninitialized or runtime-initialized global variables into the
1298 common segment, as C does. This saves space in the executable at the
1299 cost of not diagnosing duplicate definitions. If you compile with this
1300 flag and your program mysteriously crashes after @code{main()} has
1301 completed, you may have an object that is being destroyed twice because
1302 two definitions were merged.
1304 This option is no longer useful on most targets, now that support has
1305 been added for putting variables into BSS without making them common.
1307 @item -fno-const-strings
1308 @opindex fno-const-strings
1309 Give string constants type @code{char *} instead of type @code{const
1310 char *}. By default, G++ uses type @code{const char *} as required by
1311 the standard. Even if you use @option{-fno-const-strings}, you cannot
1312 actually modify the value of a string constant, unless you also use
1313 @option{-fwritable-strings}.
1315 This option might be removed in a future release of G++. For maximum
1316 portability, you should structure your code so that it works with
1317 string constants that have type @code{const char *}.
1319 @item -fno-elide-constructors
1320 @opindex fno-elide-constructors
1321 The C++ standard allows an implementation to omit creating a temporary
1322 which is only used to initialize another object of the same type.
1323 Specifying this option disables that optimization, and forces G++ to
1324 call the copy constructor in all cases.
1326 @item -fno-enforce-eh-specs
1327 @opindex fno-enforce-eh-specs
1328 Don't check for violation of exception specifications at runtime. This
1329 option violates the C++ standard, but may be useful for reducing code
1330 size in production builds, much like defining @samp{NDEBUG}. The compiler
1331 will still optimize based on the exception specifications.
1333 @item -fexternal-templates
1334 @opindex fexternal-templates
1336 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1337 template instantiation; template instances are emitted or not according
1338 to the location of the template definition. @xref{Template
1339 Instantiation}, for more information.
1341 This option is deprecated.
1343 @item -falt-external-templates
1344 @opindex falt-external-templates
1345 Similar to @option{-fexternal-templates}, but template instances are
1346 emitted or not according to the place where they are first instantiated.
1347 @xref{Template Instantiation}, for more information.
1349 This option is deprecated.
1352 @itemx -fno-for-scope
1354 @opindex fno-for-scope
1355 If @option{-ffor-scope} is specified, the scope of variables declared in
1356 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1357 as specified by the C++ standard.
1358 If @option{-fno-for-scope} is specified, the scope of variables declared in
1359 a @i{for-init-statement} extends to the end of the enclosing scope,
1360 as was the case in old versions of G++, and other (traditional)
1361 implementations of C++.
1363 The default if neither flag is given to follow the standard,
1364 but to allow and give a warning for old-style code that would
1365 otherwise be invalid, or have different behavior.
1367 @item -fno-gnu-keywords
1368 @opindex fno-gnu-keywords
1369 Do not recognize @code{typeof} as a keyword, so that code can use this
1370 word as an identifier. You can use the keyword @code{__typeof__} instead.
1371 @option{-ansi} implies @option{-fno-gnu-keywords}.
1373 @item -fno-implicit-templates
1374 @opindex fno-implicit-templates
1375 Never emit code for non-inline templates which are instantiated
1376 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1377 @xref{Template Instantiation}, for more information.
1379 @item -fno-implicit-inline-templates
1380 @opindex fno-implicit-inline-templates
1381 Don't emit code for implicit instantiations of inline templates, either.
1382 The default is to handle inlines differently so that compiles with and
1383 without optimization will need the same set of explicit instantiations.
1385 @item -fno-implement-inlines
1386 @opindex fno-implement-inlines
1387 To save space, do not emit out-of-line copies of inline functions
1388 controlled by @samp{#pragma implementation}. This will cause linker
1389 errors if these functions are not inlined everywhere they are called.
1391 @item -fms-extensions
1392 @opindex fms-extensions
1393 Disable pedantic warnings about constructs used in MFC, such as implicit
1394 int and getting a pointer to member function via non-standard syntax.
1396 @item -fno-nonansi-builtins
1397 @opindex fno-nonansi-builtins
1398 Disable built-in declarations of functions that are not mandated by
1399 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1400 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1402 @item -fno-operator-names
1403 @opindex fno-operator-names
1404 Do not treat the operator name keywords @code{and}, @code{bitand},
1405 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1406 synonyms as keywords.
1408 @item -fno-optional-diags
1409 @opindex fno-optional-diags
1410 Disable diagnostics that the standard says a compiler does not need to
1411 issue. Currently, the only such diagnostic issued by G++ is the one for
1412 a name having multiple meanings within a class.
1415 @opindex fpermissive
1416 Downgrade messages about nonconformant code from errors to warnings. By
1417 default, G++ effectively sets @option{-pedantic-errors} without
1418 @option{-pedantic}; this option reverses that. This behavior and this
1419 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1423 Enable automatic template instantiation at link time. This option also
1424 implies @option{-fno-implicit-templates}. @xref{Template
1425 Instantiation}, for more information.
1429 Disable generation of information about every class with virtual
1430 functions for use by the C++ runtime type identification features
1431 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1432 of the language, you can save some space by using this flag. Note that
1433 exception handling uses the same information, but it will generate it as
1438 Emit statistics about front-end processing at the end of the compilation.
1439 This information is generally only useful to the G++ development team.
1441 @item -ftemplate-depth-@var{n}
1442 @opindex ftemplate-depth
1443 Set the maximum instantiation depth for template classes to @var{n}.
1444 A limit on the template instantiation depth is needed to detect
1445 endless recursions during template class instantiation. ANSI/ISO C++
1446 conforming programs must not rely on a maximum depth greater than 17.
1448 @item -fuse-cxa-atexit
1449 @opindex fuse-cxa-atexit
1450 Register destructors for objects with static storage duration with the
1451 @code{__cxa_atexit} function rather than the @code{atexit} function.
1452 This option is required for fully standards-compliant handling of static
1453 destructors, but will only work if your C library supports
1454 @code{__cxa_atexit}.
1458 Emit special relocations for vtables and virtual function references
1459 so that the linker can identify unused virtual functions and zero out
1460 vtable slots that refer to them. This is most useful with
1461 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1462 also discard the functions themselves.
1464 This optimization requires GNU as and GNU ld. Not all systems support
1465 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1469 Do not use weak symbol support, even if it is provided by the linker.
1470 By default, G++ will use weak symbols if they are available. This
1471 option exists only for testing, and should not be used by end-users;
1472 it will result in inferior code and has no benefits. This option may
1473 be removed in a future release of G++.
1477 Do not search for header files in the standard directories specific to
1478 C++, but do still search the other standard directories. (This option
1479 is used when building the C++ library.)
1482 In addition, these optimization, warning, and code generation options
1483 have meanings only for C++ programs:
1486 @item -fno-default-inline
1487 @opindex fno-default-inline
1488 Do not assume @samp{inline} for functions defined inside a class scope.
1489 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1490 functions will have linkage like inline functions; they just won't be
1493 @item -Wabi @r{(C++ only)}
1495 Warn when G++ generates code that is probably not compatible with the
1496 vendor-neutral C++ ABI. Although an effort has been made to warn about
1497 all such cases, there are probably some cases that are not warned about,
1498 even though G++ is generating incompatible code. There may also be
1499 cases where warnings are emitted even though the code that is generated
1502 You should rewrite your code to avoid these warnings if you are
1503 concerned about the fact that code generated by G++ may not be binary
1504 compatible with code generated by other compilers.
1506 The known incompatibilities at this point include:
1511 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1512 pack data into the same byte as a base class. For example:
1515 struct A @{ virtual void f(); int f1 : 1; @};
1516 struct B : public A @{ int f2 : 1; @};
1520 In this case, G++ will place @code{B::f2} into the same byte
1521 as@code{A::f1}; other compilers will not. You can avoid this problem
1522 by explicitly padding @code{A} so that its size is a multiple of the
1523 byte size on your platform; that will cause G++ and other compilers to
1524 layout @code{B} identically.
1527 Incorrect handling of tail-padding for virtual bases. G++ does not use
1528 tail padding when laying out virtual bases. For example:
1531 struct A @{ virtual void f(); char c1; @};
1532 struct B @{ B(); char c2; @};
1533 struct C : public A, public virtual B @{@};
1537 In this case, G++ will not place @code{B} into the tail-padding for
1538 @code{A}; other compilers will. You can avoid this problem by
1539 explicitly padding @code{A} so that its size is a multiple of its
1540 alignment (ignoring virtual base classes); that will cause G++ and other
1541 compilers to layout @code{C} identically.
1544 Incorrect handling of bit-fields with declared widths greater than that
1545 of their underlying types, when the bit-fields appear in a union. For
1549 union U @{ int i : 4096; @};
1553 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1554 union too small by the number of bits in an @code{int}.
1557 Empty classes can be placed at incorrect offsets. For example:
1567 struct C : public B, public A @{@};
1571 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1572 it should be placed at offset zero. G++ mistakenly believes that the
1573 @code{A} data member of @code{B} is already at offset zero.
1576 Names of template functions whose types involve @code{typename} or
1577 template template parameters can be mangled incorrectly.
1580 template <typename Q>
1581 void f(typename Q::X) @{@}
1583 template <template <typename> class Q>
1584 void f(typename Q<int>::X) @{@}
1588 Instantiations of these templates may be mangled incorrectly.
1592 @item -Wctor-dtor-privacy @r{(C++ only)}
1593 @opindex Wctor-dtor-privacy
1594 Warn when a class seems unusable, because all the constructors or
1595 destructors in a class are private and the class has no friends or
1596 public static member functions.
1598 @item -Wnon-virtual-dtor @r{(C++ only)}
1599 @opindex Wnon-virtual-dtor
1600 Warn when a class declares a non-virtual destructor that should probably
1601 be virtual, because it looks like the class will be used polymorphically.
1602 This warning is enabled by @option{-Wall}.
1604 @item -Wreorder @r{(C++ only)}
1606 @cindex reordering, warning
1607 @cindex warning for reordering of member initializers
1608 Warn when the order of member initializers given in the code does not
1609 match the order in which they must be executed. For instance:
1615 A(): j (0), i (1) @{ @}
1619 Here the compiler will warn that the member initializers for @samp{i}
1620 and @samp{j} will be rearranged to match the declaration order of the
1621 members. This warning is enabled by @option{-Wall}.
1624 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1627 @item -Weffc++ @r{(C++ only)}
1629 Warn about violations of the following style guidelines from Scott Meyers'
1630 @cite{Effective C++} book:
1634 Item 11: Define a copy constructor and an assignment operator for classes
1635 with dynamically allocated memory.
1638 Item 12: Prefer initialization to assignment in constructors.
1641 Item 14: Make destructors virtual in base classes.
1644 Item 15: Have @code{operator=} return a reference to @code{*this}.
1647 Item 23: Don't try to return a reference when you must return an object.
1651 and about violations of the following style guidelines from Scott Meyers'
1652 @cite{More Effective C++} book:
1656 Item 6: Distinguish between prefix and postfix forms of increment and
1657 decrement operators.
1660 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1664 If you use this option, you should be aware that the standard library
1665 headers do not obey all of these guidelines; you can use @samp{grep -v}
1666 to filter out those warnings.
1668 @item -Wno-deprecated @r{(C++ only)}
1669 @opindex Wno-deprecated
1670 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1672 @item -Wno-non-template-friend @r{(C++ only)}
1673 @opindex Wno-non-template-friend
1674 Disable warnings when non-templatized friend functions are declared
1675 within a template. With the advent of explicit template specification
1676 support in G++, if the name of the friend is an unqualified-id (i.e.,
1677 @samp{friend foo(int)}), the C++ language specification demands that the
1678 friend declare or define an ordinary, nontemplate function. (Section
1679 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1680 could be interpreted as a particular specialization of a templatized
1681 function. Because this non-conforming behavior is no longer the default
1682 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1683 check existing code for potential trouble spots, and is on by default.
1684 This new compiler behavior can be turned off with
1685 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1686 but disables the helpful warning.
1688 @item -Wold-style-cast @r{(C++ only)}
1689 @opindex Wold-style-cast
1690 Warn if an old-style (C-style) cast to a non-void type is used within
1691 a C++ program. The new-style casts (@samp{static_cast},
1692 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1693 unintended effects, and much easier to grep for.
1695 @item -Woverloaded-virtual @r{(C++ only)}
1696 @opindex Woverloaded-virtual
1697 @cindex overloaded virtual fn, warning
1698 @cindex warning for overloaded virtual fn
1699 Warn when a function declaration hides virtual functions from a
1700 base class. For example, in:
1707 struct B: public A @{
1712 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1720 will fail to compile.
1722 @item -Wno-pmf-conversions @r{(C++ only)}
1723 @opindex Wno-pmf-conversions
1724 Disable the diagnostic for converting a bound pointer to member function
1727 @item -Wsign-promo @r{(C++ only)}
1728 @opindex Wsign-promo
1729 Warn when overload resolution chooses a promotion from unsigned or
1730 enumeral type to a signed type over a conversion to an unsigned type of
1731 the same size. Previous versions of G++ would try to preserve
1732 unsignedness, but the standard mandates the current behavior.
1734 @item -Wsynth @r{(C++ only)}
1736 @cindex warning for synthesized methods
1737 @cindex synthesized methods, warning
1738 Warn when G++'s synthesis behavior does not match that of cfront. For
1744 A& operator = (int);
1754 In this example, G++ will synthesize a default @samp{A& operator =
1755 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1758 @node Objective-C Dialect Options
1759 @section Options Controlling Objective-C Dialect
1761 @cindex compiler options, Objective-C
1762 @cindex Objective-C options, command line
1763 @cindex options, Objective-C
1764 This section describes the command-line options that are only meaningful
1765 for Objective-C programs; but you can also use most of the GNU compiler
1766 options regardless of what language your program is in. For example,
1767 you might compile a file @code{some_class.m} like this:
1770 gcc -g -fgnu-runtime -O -c some_class.m
1774 In this example, only @option{-fgnu-runtime} is an option meant only for
1775 Objective-C programs; you can use the other options with any language
1778 Here is a list of options that are @emph{only} for compiling Objective-C
1782 @item -fconstant-string-class=@var{class-name}
1783 @opindex fconstant-string-class
1784 Use @var{class-name} as the name of the class to instantiate for each
1785 literal string specified with the syntax @code{@@"@dots{}"}. The default
1786 class name is @code{NXConstantString}.
1789 @opindex fgnu-runtime
1790 Generate object code compatible with the standard GNU Objective-C
1791 runtime. This is the default for most types of systems.
1793 @item -fnext-runtime
1794 @opindex fnext-runtime
1795 Generate output compatible with the NeXT runtime. This is the default
1796 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1797 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1802 Dump interface declarations for all classes seen in the source file to a
1803 file named @file{@var{sourcename}.decl}.
1806 @opindex Wno-protocol
1807 If a class is declared to implement a protocol, a warning is issued for
1808 every method in the protocol that is not implemented by the class. The
1809 default behavior is to issue a warning for every method not explicitly
1810 implemented in the class, even if a method implementation is inherited
1811 from the superclass. If you use the @code{-Wno-protocol} option, then
1812 methods inherited from the superclass are considered to be implemented,
1813 and no warning is issued for them.
1817 Warn if multiple methods of different types for the same selector are
1818 found during compilation. The check is performed on the list of methods
1819 in the final stage of compilation. Additionally, a check is performed
1820 that for each selector appearing in a @code{@@selector(@dots{})}
1821 expression, a corresponding method with that selector has been found
1822 during compilation. Because these checks scan the method table only at
1823 the end of compilation, these warnings are not produced if the final
1824 stage of compilation is not reached, for example because an error is
1825 found during compilation, or because the @code{-fsyntax-only} option is
1828 @item -Wundeclared-selector
1829 @opindex Wundeclared-selector
1830 Warn if a @code{@@selector(@dots{})} expression referring to an
1831 undeclared selector is found. A selector is considered undeclared if no
1832 method with that name has been declared (explicitly, in an
1833 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1834 an @code{@@implementation} section) before the
1835 @code{@@selector(@dots{})} expression. This option always performs its
1836 checks as soon as a @code{@@selector(@dots{})} expression is found
1837 (while @code{-Wselector} only performs its checks in the final stage of
1838 compilation), and so additionally enforces the coding style convention
1839 that methods and selectors must be declared before being used.
1841 @c not documented because only avail via -Wp
1842 @c @item -print-objc-runtime-info
1846 @node Language Independent Options
1847 @section Options to Control Diagnostic Messages Formatting
1848 @cindex options to control diagnostics formatting
1849 @cindex diagnostic messages
1850 @cindex message formatting
1852 Traditionally, diagnostic messages have been formatted irrespective of
1853 the output device's aspect (e.g.@: its width, @dots{}). The options described
1854 below can be used to control the diagnostic messages formatting
1855 algorithm, e.g.@: how many characters per line, how often source location
1856 information should be reported. Right now, only the C++ front end can
1857 honor these options. However it is expected, in the near future, that
1858 the remaining front ends would be able to digest them correctly.
1861 @item -fmessage-length=@var{n}
1862 @opindex fmessage-length
1863 Try to format error messages so that they fit on lines of about @var{n}
1864 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1865 the front ends supported by GCC@. If @var{n} is zero, then no
1866 line-wrapping will be done; each error message will appear on a single
1869 @opindex fdiagnostics-show-location
1870 @item -fdiagnostics-show-location=once
1871 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1872 reporter to emit @emph{once} source location information; that is, in
1873 case the message is too long to fit on a single physical line and has to
1874 be wrapped, the source location won't be emitted (as prefix) again,
1875 over and over, in subsequent continuation lines. This is the default
1878 @item -fdiagnostics-show-location=every-line
1879 Only meaningful in line-wrapping mode. Instructs the diagnostic
1880 messages reporter to emit the same source location information (as
1881 prefix) for physical lines that result from the process of breaking
1882 a message which is too long to fit on a single line.
1886 @node Warning Options
1887 @section Options to Request or Suppress Warnings
1888 @cindex options to control warnings
1889 @cindex warning messages
1890 @cindex messages, warning
1891 @cindex suppressing warnings
1893 Warnings are diagnostic messages that report constructions which
1894 are not inherently erroneous but which are risky or suggest there
1895 may have been an error.
1897 You can request many specific warnings with options beginning @samp{-W},
1898 for example @option{-Wimplicit} to request warnings on implicit
1899 declarations. Each of these specific warning options also has a
1900 negative form beginning @samp{-Wno-} to turn off warnings;
1901 for example, @option{-Wno-implicit}. This manual lists only one of the
1902 two forms, whichever is not the default.
1904 The following options control the amount and kinds of warnings produced
1905 by GCC; for further, language-specific options also refer to
1906 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1909 @cindex syntax checking
1911 @opindex fsyntax-only
1912 Check the code for syntax errors, but don't do anything beyond that.
1916 Issue all the warnings demanded by strict ISO C and ISO C++;
1917 reject all programs that use forbidden extensions, and some other
1918 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1919 version of the ISO C standard specified by any @option{-std} option used.
1921 Valid ISO C and ISO C++ programs should compile properly with or without
1922 this option (though a rare few will require @option{-ansi} or a
1923 @option{-std} option specifying the required version of ISO C)@. However,
1924 without this option, certain GNU extensions and traditional C and C++
1925 features are supported as well. With this option, they are rejected.
1927 @option{-pedantic} does not cause warning messages for use of the
1928 alternate keywords whose names begin and end with @samp{__}. Pedantic
1929 warnings are also disabled in the expression that follows
1930 @code{__extension__}. However, only system header files should use
1931 these escape routes; application programs should avoid them.
1932 @xref{Alternate Keywords}.
1934 Some users try to use @option{-pedantic} to check programs for strict ISO
1935 C conformance. They soon find that it does not do quite what they want:
1936 it finds some non-ISO practices, but not all---only those for which
1937 ISO C @emph{requires} a diagnostic, and some others for which
1938 diagnostics have been added.
1940 A feature to report any failure to conform to ISO C might be useful in
1941 some instances, but would require considerable additional work and would
1942 be quite different from @option{-pedantic}. We don't have plans to
1943 support such a feature in the near future.
1945 Where the standard specified with @option{-std} represents a GNU
1946 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1947 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1948 extended dialect is based. Warnings from @option{-pedantic} are given
1949 where they are required by the base standard. (It would not make sense
1950 for such warnings to be given only for features not in the specified GNU
1951 C dialect, since by definition the GNU dialects of C include all
1952 features the compiler supports with the given option, and there would be
1953 nothing to warn about.)
1955 @item -pedantic-errors
1956 @opindex pedantic-errors
1957 Like @option{-pedantic}, except that errors are produced rather than
1962 Inhibit all warning messages.
1966 Inhibit warning messages about the use of @samp{#import}.
1968 @item -Wchar-subscripts
1969 @opindex Wchar-subscripts
1970 Warn if an array subscript has type @code{char}. This is a common cause
1971 of error, as programmers often forget that this type is signed on some
1976 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1977 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1981 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1982 the arguments supplied have types appropriate to the format string
1983 specified, and that the conversions specified in the format string make
1984 sense. This includes standard functions, and others specified by format
1985 attributes (@pxref{Function Attributes}), in the @code{printf},
1986 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1987 not in the C standard) families.
1989 The formats are checked against the format features supported by GNU
1990 libc version 2.2. These include all ISO C90 and C99 features, as well
1991 as features from the Single Unix Specification and some BSD and GNU
1992 extensions. Other library implementations may not support all these
1993 features; GCC does not support warning about features that go beyond a
1994 particular library's limitations. However, if @option{-pedantic} is used
1995 with @option{-Wformat}, warnings will be given about format features not
1996 in the selected standard version (but not for @code{strfmon} formats,
1997 since those are not in any version of the C standard). @xref{C Dialect
1998 Options,,Options Controlling C Dialect}.
2000 Since @option{-Wformat} also checks for null format arguments for
2001 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2003 @option{-Wformat} is included in @option{-Wall}. For more control over some
2004 aspects of format checking, the options @option{-Wno-format-y2k},
2005 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2006 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2007 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2009 @item -Wno-format-y2k
2010 @opindex Wno-format-y2k
2011 If @option{-Wformat} is specified, do not warn about @code{strftime}
2012 formats which may yield only a two-digit year.
2014 @item -Wno-format-extra-args
2015 @opindex Wno-format-extra-args
2016 If @option{-Wformat} is specified, do not warn about excess arguments to a
2017 @code{printf} or @code{scanf} format function. The C standard specifies
2018 that such arguments are ignored.
2020 Where the unused arguments lie between used arguments that are
2021 specified with @samp{$} operand number specifications, normally
2022 warnings are still given, since the implementation could not know what
2023 type to pass to @code{va_arg} to skip the unused arguments. However,
2024 in the case of @code{scanf} formats, this option will suppress the
2025 warning if the unused arguments are all pointers, since the Single
2026 Unix Specification says that such unused arguments are allowed.
2028 @item -Wno-format-zero-length
2029 @opindex Wno-format-zero-length
2030 If @option{-Wformat} is specified, do not warn about zero-length formats.
2031 The C standard specifies that zero-length formats are allowed.
2033 @item -Wformat-nonliteral
2034 @opindex Wformat-nonliteral
2035 If @option{-Wformat} is specified, also warn if the format string is not a
2036 string literal and so cannot be checked, unless the format function
2037 takes its format arguments as a @code{va_list}.
2039 @item -Wformat-security
2040 @opindex Wformat-security
2041 If @option{-Wformat} is specified, also warn about uses of format
2042 functions that represent possible security problems. At present, this
2043 warns about calls to @code{printf} and @code{scanf} functions where the
2044 format string is not a string literal and there are no format arguments,
2045 as in @code{printf (foo);}. This may be a security hole if the format
2046 string came from untrusted input and contains @samp{%n}. (This is
2047 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2048 in future warnings may be added to @option{-Wformat-security} that are not
2049 included in @option{-Wformat-nonliteral}.)
2053 Enable @option{-Wformat} plus format checks not included in
2054 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2055 -Wformat-nonliteral -Wformat-security}.
2059 Enable warning about passing a null pointer for arguments marked as
2060 requiring a non-null value by the @code{nonnull} function attribute.
2062 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2063 can be disabled with the @option{-Wno-nonnull} option.
2065 @item -Wimplicit-int
2066 @opindex Wimplicit-int
2067 Warn when a declaration does not specify a type.
2069 @item -Wimplicit-function-declaration
2070 @itemx -Werror-implicit-function-declaration
2071 @opindex Wimplicit-function-declaration
2072 @opindex Werror-implicit-function-declaration
2073 Give a warning (or error) whenever a function is used before being
2078 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2082 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2083 function with external linkage, returning int, taking either zero
2084 arguments, two, or three arguments of appropriate types.
2086 @item -Wmissing-braces
2087 @opindex Wmissing-braces
2088 Warn if an aggregate or union initializer is not fully bracketed. In
2089 the following example, the initializer for @samp{a} is not fully
2090 bracketed, but that for @samp{b} is fully bracketed.
2093 int a[2][2] = @{ 0, 1, 2, 3 @};
2094 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2098 @opindex Wparentheses
2099 Warn if parentheses are omitted in certain contexts, such
2100 as when there is an assignment in a context where a truth value
2101 is expected, or when operators are nested whose precedence people
2102 often get confused about.
2104 Also warn about constructions where there may be confusion to which
2105 @code{if} statement an @code{else} branch belongs. Here is an example of
2120 In C, every @code{else} branch belongs to the innermost possible @code{if}
2121 statement, which in this example is @code{if (b)}. This is often not
2122 what the programmer expected, as illustrated in the above example by
2123 indentation the programmer chose. When there is the potential for this
2124 confusion, GCC will issue a warning when this flag is specified.
2125 To eliminate the warning, add explicit braces around the innermost
2126 @code{if} statement so there is no way the @code{else} could belong to
2127 the enclosing @code{if}. The resulting code would look like this:
2143 @item -Wsequence-point
2144 @opindex Wsequence-point
2145 Warn about code that may have undefined semantics because of violations
2146 of sequence point rules in the C standard.
2148 The C standard defines the order in which expressions in a C program are
2149 evaluated in terms of @dfn{sequence points}, which represent a partial
2150 ordering between the execution of parts of the program: those executed
2151 before the sequence point, and those executed after it. These occur
2152 after the evaluation of a full expression (one which is not part of a
2153 larger expression), after the evaluation of the first operand of a
2154 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2155 function is called (but after the evaluation of its arguments and the
2156 expression denoting the called function), and in certain other places.
2157 Other than as expressed by the sequence point rules, the order of
2158 evaluation of subexpressions of an expression is not specified. All
2159 these rules describe only a partial order rather than a total order,
2160 since, for example, if two functions are called within one expression
2161 with no sequence point between them, the order in which the functions
2162 are called is not specified. However, the standards committee have
2163 ruled that function calls do not overlap.
2165 It is not specified when between sequence points modifications to the
2166 values of objects take effect. Programs whose behavior depends on this
2167 have undefined behavior; the C standard specifies that ``Between the
2168 previous and next sequence point an object shall have its stored value
2169 modified at most once by the evaluation of an expression. Furthermore,
2170 the prior value shall be read only to determine the value to be
2171 stored.''. If a program breaks these rules, the results on any
2172 particular implementation are entirely unpredictable.
2174 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2175 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2176 diagnosed by this option, and it may give an occasional false positive
2177 result, but in general it has been found fairly effective at detecting
2178 this sort of problem in programs.
2180 The present implementation of this option only works for C programs. A
2181 future implementation may also work for C++ programs.
2183 The C standard is worded confusingly, therefore there is some debate
2184 over the precise meaning of the sequence point rules in subtle cases.
2185 Links to discussions of the problem, including proposed formal
2186 definitions, may be found on our readings page, at
2187 @w{@uref{http://gcc.gnu.org/readings.html}}.
2190 @opindex Wreturn-type
2191 Warn whenever a function is defined with a return-type that defaults to
2192 @code{int}. Also warn about any @code{return} statement with no
2193 return-value in a function whose return-type is not @code{void}.
2195 For C++, a function without return type always produces a diagnostic
2196 message, even when @option{-Wno-return-type} is specified. The only
2197 exceptions are @samp{main} and functions defined in system headers.
2201 Warn whenever a @code{switch} statement has an index of enumeral type
2202 and lacks a @code{case} for one or more of the named codes of that
2203 enumeration. (The presence of a @code{default} label prevents this
2204 warning.) @code{case} labels outside the enumeration range also
2205 provoke warnings when this option is used.
2207 @item -Wswitch-default
2208 @opindex Wswitch-switch
2209 Warn whenever a @code{switch} statement does not have a @code{default}
2213 @opindex Wswitch-enum
2214 Warn whenever a @code{switch} statement has an index of enumeral type
2215 and lacks a @code{case} for one or more of the named codes of that
2216 enumeration. @code{case} labels outside the enumeration range also
2217 provoke warnings when this option is used.
2221 Warn if any trigraphs are encountered that might change the meaning of
2222 the program (trigraphs within comments are not warned about).
2224 @item -Wunused-function
2225 @opindex Wunused-function
2226 Warn whenever a static function is declared but not defined or a
2227 non\-inline static function is unused.
2229 @item -Wunused-label
2230 @opindex Wunused-label
2231 Warn whenever a label is declared but not used.
2233 To suppress this warning use the @samp{unused} attribute
2234 (@pxref{Variable Attributes}).
2236 @item -Wunused-parameter
2237 @opindex Wunused-parameter
2238 Warn whenever a function parameter is unused aside from its declaration.
2240 To suppress this warning use the @samp{unused} attribute
2241 (@pxref{Variable Attributes}).
2243 @item -Wunused-variable
2244 @opindex Wunused-variable
2245 Warn whenever a local variable or non-constant static variable is unused
2246 aside from its declaration
2248 To suppress this warning use the @samp{unused} attribute
2249 (@pxref{Variable Attributes}).
2251 @item -Wunused-value
2252 @opindex Wunused-value
2253 Warn whenever a statement computes a result that is explicitly not used.
2255 To suppress this warning cast the expression to @samp{void}.
2259 All the above @option{-Wunused} options combined.
2261 In order to get a warning about an unused function parameter, you must
2262 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2263 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2265 @item -Wuninitialized
2266 @opindex Wuninitialized
2267 Warn if an automatic variable is used without first being initialized or
2268 if a variable may be clobbered by a @code{setjmp} call.
2270 These warnings are possible only in optimizing compilation,
2271 because they require data flow information that is computed only
2272 when optimizing. If you don't specify @option{-O}, you simply won't
2275 These warnings occur only for variables that are candidates for
2276 register allocation. Therefore, they do not occur for a variable that
2277 is declared @code{volatile}, or whose address is taken, or whose size
2278 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2279 structures, unions or arrays, even when they are in registers.
2281 Note that there may be no warning about a variable that is used only
2282 to compute a value that itself is never used, because such
2283 computations may be deleted by data flow analysis before the warnings
2286 These warnings are made optional because GCC is not smart
2287 enough to see all the reasons why the code might be correct
2288 despite appearing to have an error. Here is one example of how
2309 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2310 always initialized, but GCC doesn't know this. Here is
2311 another common case:
2316 if (change_y) save_y = y, y = new_y;
2318 if (change_y) y = save_y;
2323 This has no bug because @code{save_y} is used only if it is set.
2325 @cindex @code{longjmp} warnings
2326 This option also warns when a non-volatile automatic variable might be
2327 changed by a call to @code{longjmp}. These warnings as well are possible
2328 only in optimizing compilation.
2330 The compiler sees only the calls to @code{setjmp}. It cannot know
2331 where @code{longjmp} will be called; in fact, a signal handler could
2332 call it at any point in the code. As a result, you may get a warning
2333 even when there is in fact no problem because @code{longjmp} cannot
2334 in fact be called at the place which would cause a problem.
2336 Some spurious warnings can be avoided if you declare all the functions
2337 you use that never return as @code{noreturn}. @xref{Function
2340 @item -Wunknown-pragmas
2341 @opindex Wunknown-pragmas
2342 @cindex warning for unknown pragmas
2343 @cindex unknown pragmas, warning
2344 @cindex pragmas, warning of unknown
2345 Warn when a #pragma directive is encountered which is not understood by
2346 GCC@. If this command line option is used, warnings will even be issued
2347 for unknown pragmas in system header files. This is not the case if
2348 the warnings were only enabled by the @option{-Wall} command line option.
2350 @item -Wstrict-aliasing
2351 @opindex Wstrict-aliasing
2352 This option is only active when @option{-fstrict-aliasing} is active.
2353 It warns about code which might break the strict aliasing rules that the
2354 compiler is using for optimization. The warning does not catch all
2355 cases, but does attempt to catch the more common pitfalls. It is
2356 included in @option{-Wall}.
2360 All of the above @samp{-W} options combined. This enables all the
2361 warnings about constructions that some users consider questionable, and
2362 that are easy to avoid (or modify to prevent the warning), even in
2363 conjunction with macros. This also enables some language-specific
2364 warnings described in @ref{C++ Dialect Options} and
2365 @ref{Objective-C Dialect Options}.
2368 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2369 Some of them warn about constructions that users generally do not
2370 consider questionable, but which occasionally you might wish to check
2371 for; others warn about constructions that are necessary or hard to avoid
2372 in some cases, and there is no simple way to modify the code to suppress
2379 (This option used to be called @option{-W}. The older name is still
2380 supported, but the newer name is more descriptive.) Print extra warning
2381 messages for these events:
2385 A function can return either with or without a value. (Falling
2386 off the end of the function body is considered returning without
2387 a value.) For example, this function would evoke such a
2401 An expression-statement or the left-hand side of a comma expression
2402 contains no side effects.
2403 To suppress the warning, cast the unused expression to void.
2404 For example, an expression such as @samp{x[i,j]} will cause a warning,
2405 but @samp{x[(void)i,j]} will not.
2408 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2411 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2412 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2413 that of ordinary mathematical notation.
2416 Storage-class specifiers like @code{static} are not the first things in
2417 a declaration. According to the C Standard, this usage is obsolescent.
2420 The return type of a function has a type qualifier such as @code{const}.
2421 Such a type qualifier has no effect, since the value returned by a
2422 function is not an lvalue. (But don't warn about the GNU extension of
2423 @code{volatile void} return types. That extension will be warned about
2424 if @option{-pedantic} is specified.)
2427 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2431 A comparison between signed and unsigned values could produce an
2432 incorrect result when the signed value is converted to unsigned.
2433 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2436 An aggregate has an initializer which does not initialize all members.
2437 For example, the following code would cause such a warning, because
2438 @code{x.h} would be implicitly initialized to zero:
2441 struct s @{ int f, g, h; @};
2442 struct s x = @{ 3, 4 @};
2446 A function parameter is declared without a type specifier in K&R-style
2454 An empty body occurs in an @samp{if} or @samp{else} statement.
2457 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2458 @samp{>}, or @samp{>=}.
2461 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2464 Any of several floating-point events that often indicate errors, such as
2465 overflow, underflow, loss of precision, etc.
2467 @item @r{(C++ only)}
2468 An enumerator and a non-enumerator both appear in a conditional expression.
2470 @item @r{(C++ only)}
2471 A non-static reference or non-static @samp{const} member appears in a
2472 class without constructors.
2474 @item @r{(C++ only)}
2475 Ambiguous virtual bases.
2477 @item @r{(C++ only)}
2478 Subscripting an array which has been declared @samp{register}.
2480 @item @r{(C++ only)}
2481 Taking the address of a variable which has been declared @samp{register}.
2483 @item @r{(C++ only)}
2484 A base class is not initialized in a derived class' copy constructor.
2487 @item -Wno-div-by-zero
2488 @opindex Wno-div-by-zero
2489 @opindex Wdiv-by-zero
2490 Do not warn about compile-time integer division by zero. Floating point
2491 division by zero is not warned about, as it can be a legitimate way of
2492 obtaining infinities and NaNs.
2494 @item -Wsystem-headers
2495 @opindex Wsystem-headers
2496 @cindex warnings from system headers
2497 @cindex system headers, warnings from
2498 Print warning messages for constructs found in system header files.
2499 Warnings from system headers are normally suppressed, on the assumption
2500 that they usually do not indicate real problems and would only make the
2501 compiler output harder to read. Using this command line option tells
2502 GCC to emit warnings from system headers as if they occurred in user
2503 code. However, note that using @option{-Wall} in conjunction with this
2504 option will @emph{not} warn about unknown pragmas in system
2505 headers---for that, @option{-Wunknown-pragmas} must also be used.
2508 @opindex Wfloat-equal
2509 Warn if floating point values are used in equality comparisons.
2511 The idea behind this is that sometimes it is convenient (for the
2512 programmer) to consider floating-point values as approximations to
2513 infinitely precise real numbers. If you are doing this, then you need
2514 to compute (by analyzing the code, or in some other way) the maximum or
2515 likely maximum error that the computation introduces, and allow for it
2516 when performing comparisons (and when producing output, but that's a
2517 different problem). In particular, instead of testing for equality, you
2518 would check to see whether the two values have ranges that overlap; and
2519 this is done with the relational operators, so equality comparisons are
2522 @item -Wtraditional @r{(C only)}
2523 @opindex Wtraditional
2524 Warn about certain constructs that behave differently in traditional and
2525 ISO C@. Also warn about ISO C constructs that have no traditional C
2526 equivalent, and/or problematic constructs which should be avoided.
2530 Macro parameters that appear within string literals in the macro body.
2531 In traditional C macro replacement takes place within string literals,
2532 but does not in ISO C@.
2535 In traditional C, some preprocessor directives did not exist.
2536 Traditional preprocessors would only consider a line to be a directive
2537 if the @samp{#} appeared in column 1 on the line. Therefore
2538 @option{-Wtraditional} warns about directives that traditional C
2539 understands but would ignore because the @samp{#} does not appear as the
2540 first character on the line. It also suggests you hide directives like
2541 @samp{#pragma} not understood by traditional C by indenting them. Some
2542 traditional implementations would not recognize @samp{#elif}, so it
2543 suggests avoiding it altogether.
2546 A function-like macro that appears without arguments.
2549 The unary plus operator.
2552 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2553 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2554 constants.) Note, these suffixes appear in macros defined in the system
2555 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2556 Use of these macros in user code might normally lead to spurious
2557 warnings, however gcc's integrated preprocessor has enough context to
2558 avoid warning in these cases.
2561 A function declared external in one block and then used after the end of
2565 A @code{switch} statement has an operand of type @code{long}.
2568 A non-@code{static} function declaration follows a @code{static} one.
2569 This construct is not accepted by some traditional C compilers.
2572 The ISO type of an integer constant has a different width or
2573 signedness from its traditional type. This warning is only issued if
2574 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2575 typically represent bit patterns, are not warned about.
2578 Usage of ISO string concatenation is detected.
2581 Initialization of automatic aggregates.
2584 Identifier conflicts with labels. Traditional C lacks a separate
2585 namespace for labels.
2588 Initialization of unions. If the initializer is zero, the warning is
2589 omitted. This is done under the assumption that the zero initializer in
2590 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2591 initializer warnings and relies on default initialization to zero in the
2595 Conversions by prototypes between fixed/floating point values and vice
2596 versa. The absence of these prototypes when compiling with traditional
2597 C would cause serious problems. This is a subset of the possible
2598 conversion warnings, for the full set use @option{-Wconversion}.
2601 Use of ISO C style function definitions. This warning intentionally is
2602 @emph{not} issued for prototype declarations or variadic functions
2603 because these ISO C features will appear in your code when using
2604 libiberty's traditional C compatibility macros, @code{PARAMS} and
2605 @code{VPARAMS}. This warning is also bypassed for nested functions
2606 because that feature is already a gcc extension and thus not relevant to
2607 traditional C compatibility.
2612 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2614 @item -Wendif-labels
2615 @opindex Wendif-labels
2616 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2620 Warn whenever a local variable shadows another local variable, parameter or
2621 global variable or whenever a built-in function is shadowed.
2623 @item -Wlarger-than-@var{len}
2624 @opindex Wlarger-than
2625 Warn whenever an object of larger than @var{len} bytes is defined.
2627 @item -Wpointer-arith
2628 @opindex Wpointer-arith
2629 Warn about anything that depends on the ``size of'' a function type or
2630 of @code{void}. GNU C assigns these types a size of 1, for
2631 convenience in calculations with @code{void *} pointers and pointers
2634 @item -Wbad-function-cast @r{(C only)}
2635 @opindex Wbad-function-cast
2636 Warn whenever a function call is cast to a non-matching type.
2637 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2641 Warn whenever a pointer is cast so as to remove a type qualifier from
2642 the target type. For example, warn if a @code{const char *} is cast
2643 to an ordinary @code{char *}.
2646 @opindex Wcast-align
2647 Warn whenever a pointer is cast such that the required alignment of the
2648 target is increased. For example, warn if a @code{char *} is cast to
2649 an @code{int *} on machines where integers can only be accessed at
2650 two- or four-byte boundaries.
2652 @item -Wwrite-strings
2653 @opindex Wwrite-strings
2654 When compiling C, give string constants the type @code{const
2655 char[@var{length}]} so that
2656 copying the address of one into a non-@code{const} @code{char *}
2657 pointer will get a warning; when compiling C++, warn about the
2658 deprecated conversion from string constants to @code{char *}.
2659 These warnings will help you find at
2660 compile time code that can try to write into a string constant, but
2661 only if you have been very careful about using @code{const} in
2662 declarations and prototypes. Otherwise, it will just be a nuisance;
2663 this is why we did not make @option{-Wall} request these warnings.
2666 @opindex Wconversion
2667 Warn if a prototype causes a type conversion that is different from what
2668 would happen to the same argument in the absence of a prototype. This
2669 includes conversions of fixed point to floating and vice versa, and
2670 conversions changing the width or signedness of a fixed point argument
2671 except when the same as the default promotion.
2673 Also, warn if a negative integer constant expression is implicitly
2674 converted to an unsigned type. For example, warn about the assignment
2675 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2676 casts like @code{(unsigned) -1}.
2678 @item -Wsign-compare
2679 @opindex Wsign-compare
2680 @cindex warning for comparison of signed and unsigned values
2681 @cindex comparison of signed and unsigned values, warning
2682 @cindex signed and unsigned values, comparison warning
2683 Warn when a comparison between signed and unsigned values could produce
2684 an incorrect result when the signed value is converted to unsigned.
2685 This warning is also enabled by @option{-Wextra}; to get the other warnings
2686 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2688 @item -Waggregate-return
2689 @opindex Waggregate-return
2690 Warn if any functions that return structures or unions are defined or
2691 called. (In languages where you can return an array, this also elicits
2694 @item -Wstrict-prototypes @r{(C only)}
2695 @opindex Wstrict-prototypes
2696 Warn if a function is declared or defined without specifying the
2697 argument types. (An old-style function definition is permitted without
2698 a warning if preceded by a declaration which specifies the argument
2701 @item -Wmissing-prototypes @r{(C only)}
2702 @opindex Wmissing-prototypes
2703 Warn if a global function is defined without a previous prototype
2704 declaration. This warning is issued even if the definition itself
2705 provides a prototype. The aim is to detect global functions that fail
2706 to be declared in header files.
2708 @item -Wmissing-declarations @r{(C only)}
2709 @opindex Wmissing-declarations
2710 Warn if a global function is defined without a previous declaration.
2711 Do so even if the definition itself provides a prototype.
2712 Use this option to detect global functions that are not declared in
2715 @item -Wmissing-noreturn
2716 @opindex Wmissing-noreturn
2717 Warn about functions which might be candidates for attribute @code{noreturn}.
2718 Note these are only possible candidates, not absolute ones. Care should
2719 be taken to manually verify functions actually do not ever return before
2720 adding the @code{noreturn} attribute, otherwise subtle code generation
2721 bugs could be introduced. You will not get a warning for @code{main} in
2722 hosted C environments.
2724 @item -Wmissing-format-attribute
2725 @opindex Wmissing-format-attribute
2727 If @option{-Wformat} is enabled, also warn about functions which might be
2728 candidates for @code{format} attributes. Note these are only possible
2729 candidates, not absolute ones. GCC will guess that @code{format}
2730 attributes might be appropriate for any function that calls a function
2731 like @code{vprintf} or @code{vscanf}, but this might not always be the
2732 case, and some functions for which @code{format} attributes are
2733 appropriate may not be detected. This option has no effect unless
2734 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2736 @item -Wno-multichar
2737 @opindex Wno-multichar
2739 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2740 Usually they indicate a typo in the user's code, as they have
2741 implementation-defined values, and should not be used in portable code.
2743 @item -Wno-deprecated-declarations
2744 @opindex Wno-deprecated-declarations
2745 Do not warn about uses of functions, variables, and types marked as
2746 deprecated by using the @code{deprecated} attribute.
2747 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2748 @pxref{Type Attributes}.)
2752 Warn if a structure is given the packed attribute, but the packed
2753 attribute has no effect on the layout or size of the structure.
2754 Such structures may be mis-aligned for little benefit. For
2755 instance, in this code, the variable @code{f.x} in @code{struct bar}
2756 will be misaligned even though @code{struct bar} does not itself
2757 have the packed attribute:
2764 @} __attribute__((packed));
2774 Warn if padding is included in a structure, either to align an element
2775 of the structure or to align the whole structure. Sometimes when this
2776 happens it is possible to rearrange the fields of the structure to
2777 reduce the padding and so make the structure smaller.
2779 @item -Wredundant-decls
2780 @opindex Wredundant-decls
2781 Warn if anything is declared more than once in the same scope, even in
2782 cases where multiple declaration is valid and changes nothing.
2784 @item -Wnested-externs @r{(C only)}
2785 @opindex Wnested-externs
2786 Warn if an @code{extern} declaration is encountered within a function.
2788 @item -Wunreachable-code
2789 @opindex Wunreachable-code
2790 Warn if the compiler detects that code will never be executed.
2792 This option is intended to warn when the compiler detects that at
2793 least a whole line of source code will never be executed, because
2794 some condition is never satisfied or because it is after a
2795 procedure that never returns.
2797 It is possible for this option to produce a warning even though there
2798 are circumstances under which part of the affected line can be executed,
2799 so care should be taken when removing apparently-unreachable code.
2801 For instance, when a function is inlined, a warning may mean that the
2802 line is unreachable in only one inlined copy of the function.
2804 This option is not made part of @option{-Wall} because in a debugging
2805 version of a program there is often substantial code which checks
2806 correct functioning of the program and is, hopefully, unreachable
2807 because the program does work. Another common use of unreachable
2808 code is to provide behavior which is selectable at compile-time.
2812 Warn if a function can not be inlined and it was declared as inline.
2813 Even with this option, the compiler will not warn about failures to
2814 inline functions declared in system headers.
2816 The compiler uses a variety of heuristics to determine whether or not
2817 to inline a function. For example, the compiler takes into account
2818 the size of the function being inlined and the the amount of inlining
2819 that has already been done in the current function. Therefore,
2820 seemingly insignificant changes in the source program can cause the
2821 warnings produced by @option{-Winline} to appear or disappear.
2823 @item -Wno-invalid-offsetof @r{(C++ only)}
2824 @opindex Wno-invalid-offsetof
2825 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2826 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2827 to a non-POD type is undefined. In existing C++ implementations,
2828 however, @samp{offsetof} typically gives meaningful results even when
2829 applied to certain kinds of non-POD types. (Such as a simple
2830 @samp{struct} that fails to be a POD type only by virtue of having a
2831 constructor.) This flag is for users who are aware that they are
2832 writing nonportable code and who have deliberately chosen to ignore the
2835 The restrictions on @samp{offsetof} may be relaxed in a future version
2836 of the C++ standard.
2839 @opindex Winvalid-pch
2840 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2841 the search path but can't be used.
2845 @opindex Wno-long-long
2846 Warn if @samp{long long} type is used. This is default. To inhibit
2847 the warning messages, use @option{-Wno-long-long}. Flags
2848 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2849 only when @option{-pedantic} flag is used.
2851 @item -Wdisabled-optimization
2852 @opindex Wdisabled-optimization
2853 Warn if a requested optimization pass is disabled. This warning does
2854 not generally indicate that there is anything wrong with your code; it
2855 merely indicates that GCC's optimizers were unable to handle the code
2856 effectively. Often, the problem is that your code is too big or too
2857 complex; GCC will refuse to optimize programs when the optimization
2858 itself is likely to take inordinate amounts of time.
2862 Make all warnings into errors.
2865 @node Debugging Options
2866 @section Options for Debugging Your Program or GCC
2867 @cindex options, debugging
2868 @cindex debugging information options
2870 GCC has various special options that are used for debugging
2871 either your program or GCC:
2876 Produce debugging information in the operating system's native format
2877 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2880 On most systems that use stabs format, @option{-g} enables use of extra
2881 debugging information that only GDB can use; this extra information
2882 makes debugging work better in GDB but will probably make other debuggers
2884 refuse to read the program. If you want to control for certain whether
2885 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2886 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2887 or @option{-gvms} (see below).
2889 Unlike most other C compilers, GCC allows you to use @option{-g} with
2890 @option{-O}. The shortcuts taken by optimized code may occasionally
2891 produce surprising results: some variables you declared may not exist
2892 at all; flow of control may briefly move where you did not expect it;
2893 some statements may not be executed because they compute constant
2894 results or their values were already at hand; some statements may
2895 execute in different places because they were moved out of loops.
2897 Nevertheless it proves possible to debug optimized output. This makes
2898 it reasonable to use the optimizer for programs that might have bugs.
2900 The following options are useful when GCC is generated with the
2901 capability for more than one debugging format.
2905 Produce debugging information for use by GDB@. This means to use the
2906 most expressive format available (DWARF 2, stabs, or the native format
2907 if neither of those are supported), including GDB extensions if at all
2912 Produce debugging information in stabs format (if that is supported),
2913 without GDB extensions. This is the format used by DBX on most BSD
2914 systems. On MIPS, Alpha and System V Release 4 systems this option
2915 produces stabs debugging output which is not understood by DBX or SDB@.
2916 On System V Release 4 systems this option requires the GNU assembler.
2918 @item -feliminate-unused-debug-symbols
2919 @opindex feliminate-unused-debug-symbols
2920 Produce debugging information in stabs format (if that is supported),
2921 for only symbols that are actually used.
2925 Produce debugging information in stabs format (if that is supported),
2926 using GNU extensions understood only by the GNU debugger (GDB)@. The
2927 use of these extensions is likely to make other debuggers crash or
2928 refuse to read the program.
2932 Produce debugging information in COFF format (if that is supported).
2933 This is the format used by SDB on most System V systems prior to
2938 Produce debugging information in XCOFF format (if that is supported).
2939 This is the format used by the DBX debugger on IBM RS/6000 systems.
2943 Produce debugging information in XCOFF format (if that is supported),
2944 using GNU extensions understood only by the GNU debugger (GDB)@. The
2945 use of these extensions is likely to make other debuggers crash or
2946 refuse to read the program, and may cause assemblers other than the GNU
2947 assembler (GAS) to fail with an error.
2951 Produce debugging information in DWARF version 1 format (if that is
2952 supported). This is the format used by SDB on most System V Release 4
2955 This option is deprecated.
2959 Produce debugging information in DWARF version 1 format (if that is
2960 supported), using GNU extensions understood only by the GNU debugger
2961 (GDB)@. The use of these extensions is likely to make other debuggers
2962 crash or refuse to read the program.
2964 This option is deprecated.
2968 Produce debugging information in DWARF version 2 format (if that is
2969 supported). This is the format used by DBX on IRIX 6.
2973 Produce debugging information in VMS debug format (if that is
2974 supported). This is the format used by DEBUG on VMS systems.
2977 @itemx -ggdb@var{level}
2978 @itemx -gstabs@var{level}
2979 @itemx -gcoff@var{level}
2980 @itemx -gxcoff@var{level}
2981 @itemx -gvms@var{level}
2982 Request debugging information and also use @var{level} to specify how
2983 much information. The default level is 2.
2985 Level 1 produces minimal information, enough for making backtraces in
2986 parts of the program that you don't plan to debug. This includes
2987 descriptions of functions and external variables, but no information
2988 about local variables and no line numbers.
2990 Level 3 includes extra information, such as all the macro definitions
2991 present in the program. Some debuggers support macro expansion when
2992 you use @option{-g3}.
2994 Note that in order to avoid confusion between DWARF1 debug level 2,
2995 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2996 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2997 option to change the debug level for DWARF1 or DWARF2.
2999 @item -feliminate-dwarf2-dups
3000 @opindex feliminate-dwarf2-dups
3001 Compress DWARF2 debugging information by eliminating duplicated
3002 information about each symbol. This option only makes sense when
3003 generating DWARF2 debugging information with @option{-gdwarf-2}.
3005 @cindex @command{prof}
3008 Generate extra code to write profile information suitable for the
3009 analysis program @command{prof}. You must use this option when compiling
3010 the source files you want data about, and you must also use it when
3013 @cindex @command{gprof}
3016 Generate extra code to write profile information suitable for the
3017 analysis program @command{gprof}. You must use this option when compiling
3018 the source files you want data about, and you must also use it when
3023 Makes the compiler print out each function name as it is compiled, and
3024 print some statistics about each pass when it finishes.
3027 @opindex ftime-report
3028 Makes the compiler print some statistics about the time consumed by each
3029 pass when it finishes.
3032 @opindex fmem-report
3033 Makes the compiler print some statistics about permanent memory
3034 allocation when it finishes.
3036 @item -fprofile-arcs
3037 @opindex fprofile-arcs
3038 Add code so that program flow @dfn{arcs} are instrumented. During
3039 execution the program records how many times each branch and call is
3040 executed and how many times it is taken or returns. When the compiled
3041 program exits it saves this data to a file called
3042 @file{@var{auxname}.da} for each source file. The data may be used for
3043 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3044 test coverage analysis (@option{-ftest-coverage}). Each object file's
3045 @var{auxname} is generated from the name of the output file, if
3046 explicitly specified and it is not the final executable, otherwise it is
3047 the basename of the source file. In both cases any suffix is removed
3048 (e.g. @file{foo.da} for input file @file{dir/foo.c}, or
3049 @file{dir/foo.da} for output file specified as @option{-o dir/foo.o}).
3054 Compile the source files with @option{-fprofile-arcs} plus optimization
3055 and code generation options. For test coverage analysis, use the
3056 additional @option{-ftest-coverage} option. You do not need to profile
3057 every source file in a program.
3060 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3061 (the latter implies the former).
3064 Run the program on a representative workload to generate the arc profile
3065 information. This may be repeated any number of times. You can run
3066 concurrent instances of your program, and provided that the file system
3067 supports locking, the data files will be correctly updated. Also
3068 @code{fork} calls are detected and correctly handled (double counting
3072 For profile-directed optimizations, compile the source files again with
3073 the same optimization and code generation options plus
3074 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3075 Control Optimization}).
3078 For test coverage analysis, use @command{gcov} to produce human readable
3079 information from the @file{.bbg} and @file{.da} files. Refer to the
3080 @command{gcov} documentation for further information.
3084 With @option{-fprofile-arcs}, for each function of your program GCC
3085 creates a program flow graph, then finds a spanning tree for the graph.
3086 Only arcs that are not on the spanning tree have to be instrumented: the
3087 compiler adds code to count the number of times that these arcs are
3088 executed. When an arc is the only exit or only entrance to a block, the
3089 instrumentation code can be added to the block; otherwise, a new basic
3090 block must be created to hold the instrumentation code.
3093 @item -ftest-coverage
3094 @opindex ftest-coverage
3095 Produce a graph file that the @command{gcov} code-coverage utility
3096 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3097 show program coverage. Each source file's data file is called
3098 @file{@var{auxname}.bbg}. Refer to the @option{-fprofile-arcs} option
3099 above for a description of @var{auxname} and instructions on how to
3100 generate test coverage data. Coverage data will match the source files
3101 more closely, if you do not optimize.
3103 @item -d@var{letters}
3105 Says to make debugging dumps during compilation at times specified by
3106 @var{letters}. This is used for debugging the compiler. The file names
3107 for most of the dumps are made by appending a pass number and a word to
3108 the @var{dumpname}. @var{dumpname} is generated from the name of the
3109 output file, if explicitly specified and it is not an executable,
3110 otherwise it is the basename of the source file. In both cases any
3111 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3112 Here are the possible letters for use in @var{letters}, and their
3118 Annotate the assembler output with miscellaneous debugging information.
3121 Dump after computing branch probabilities, to @file{@var{file}.16.bp}.
3124 Dump after block reordering, to @file{@var{file}.32.bbro}.
3127 Dump after instruction combination, to the file @file{@var{file}.22.combine}.
3130 Dump after the first if conversion, to the file @file{@var{file}.17.ce1}.
3131 Also dump after the second if conversion, to the file @file{@var{file}.23.ce2}.
3134 Dump after branch target load optimization, to to @file{@var{file}.34.btl}.
3135 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3138 Dump all macro definitions, at the end of preprocessing, in addition to
3142 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
3143 @file{@var{file}.010.ussa}.
3146 Dump after the second if conversion, to @file{@var{file}.33.ce3}.
3149 Dump after control and data flow analysis, to @file{@var{file}.15.cfg}.
3150 Also dump after life analysis, to @file{@var{file}.21.life}.
3153 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.11.addressof}.
3156 Dump after global register allocation, to @file{@var{file}.27.greg}.
3159 Dump after GCSE, to @file{@var{file}.12.gcse}.
3160 Also dump after jump bypassing and control flow optimizations, to
3161 @file{@var{file}.14.bypass}.
3164 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3167 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3170 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3173 Dump after conversion from registers to stack, to @file{@var{file}.36.stack}.
3176 Dump after local register allocation, to @file{@var{file}.26.lreg}.
3179 Dump after loop optimization passes, to @file{@var{file}.13.loop} and
3180 @file{@var{file}.19.loop2}.
3183 Dump after performing the machine dependent reorganization pass, to
3184 @file{@var{file}.37.mach}.
3187 Dump after register renumbering, to @file{@var{file}.31.rnreg}.
3190 Dump after the register move pass, to @file{@var{file}.24.regmove}.
3193 Dump after post-reload optimizations, to @file{@var{file}.28.postreload}.
3196 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3199 Dump after the second scheduling pass, to @file{@var{file}.35.sched2}.
3202 Dump after CSE (including the jump optimization that sometimes follows
3203 CSE), to @file{@var{file}.019.cse}.
3206 Dump after the first scheduling pass, to @file{@var{file}.25.sched}.
3209 Dump after the second CSE pass (including the jump optimization that
3210 sometimes follows CSE), to @file{@var{file}.20.cse2}.
3213 Dump after running tracer, to @file{@var{file}.18.tracer}.
3216 Dump after null pointer elimination pass to @file{@var{file}.018.null}.
3219 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3222 Dump after the second flow pass, to @file{@var{file}.29.flow2}.
3225 Dump after SSA conditional constant propagation, to
3226 @file{@var{file}.06.ssaccp}.
3229 Dump after SSA dead code elimination, to @file{@var{file}.07.ssadce}.
3232 Dump after the peephole pass, to @file{@var{file}.30.peephole2}.
3235 Produce all the dumps listed above.
3238 Produce a core dump whenever an error occurs.
3241 Print statistics on memory usage, at the end of the run, to
3245 Annotate the assembler output with a comment indicating which
3246 pattern and alternative was used. The length of each instruction is
3250 Dump the RTL in the assembler output as a comment before each instruction.
3251 Also turns on @option{-dp} annotation.
3254 For each of the other indicated dump files (except for
3255 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3256 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3259 Just generate RTL for a function instead of compiling it. Usually used
3263 Dump debugging information during parsing, to standard error.
3266 @item -fdump-unnumbered
3267 @opindex fdump-unnumbered
3268 When doing debugging dumps (see @option{-d} option above), suppress instruction
3269 numbers and line number note output. This makes it more feasible to
3270 use diff on debugging dumps for compiler invocations with different
3271 options, in particular with and without @option{-g}.
3273 @item -fdump-translation-unit @r{(C and C++ only)}
3274 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3275 @opindex fdump-translation-unit
3276 Dump a representation of the tree structure for the entire translation
3277 unit to a file. The file name is made by appending @file{.tu} to the
3278 source file name. If the @samp{-@var{options}} form is used, @var{options}
3279 controls the details of the dump as described for the
3280 @option{-fdump-tree} options.
3282 @item -fdump-class-hierarchy @r{(C++ only)}
3283 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3284 @opindex fdump-class-hierarchy
3285 Dump a representation of each class's hierarchy and virtual function
3286 table layout to a file. The file name is made by appending @file{.class}
3287 to the source file name. If the @samp{-@var{options}} form is used,
3288 @var{options} controls the details of the dump as described for the
3289 @option{-fdump-tree} options.
3291 @item -fdump-tree-@var{switch} @r{(C++ only)}
3292 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3294 Control the dumping at various stages of processing the intermediate
3295 language tree to a file. The file name is generated by appending a switch
3296 specific suffix to the source file name. If the @samp{-@var{options}}
3297 form is used, @var{options} is a list of @samp{-} separated options that
3298 control the details of the dump. Not all options are applicable to all
3299 dumps, those which are not meaningful will be ignored. The following
3300 options are available
3304 Print the address of each node. Usually this is not meaningful as it
3305 changes according to the environment and source file. Its primary use
3306 is for tying up a dump file with a debug environment.
3308 Inhibit dumping of members of a scope or body of a function merely
3309 because that scope has been reached. Only dump such items when they
3310 are directly reachable by some other path.
3312 Turn on all options.
3315 The following tree dumps are possible:
3318 Dump before any tree based optimization, to @file{@var{file}.original}.
3320 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3322 Dump after function inlining, to @file{@var{file}.inlined}.
3325 @item -frandom-seed=@var{string}
3326 @opindex frandom-string
3327 This option provides a seed that GCC uses when it would otherwise use
3328 random numbers. It is used to generate certain symbol names
3329 that have to be different in every compiled file. It is also used to
3330 place unique stamps in coverage data files and the object files that
3331 produce them. You can use the @option{-frandom-seed} option to produce
3332 reproducibly identical object files.
3334 The @var{string} should be different for every file you compile.
3336 @item -fsched-verbose=@var{n}
3337 @opindex fsched-verbose
3338 On targets that use instruction scheduling, this option controls the
3339 amount of debugging output the scheduler prints. This information is
3340 written to standard error, unless @option{-dS} or @option{-dR} is
3341 specified, in which case it is output to the usual dump
3342 listing file, @file{.sched} or @file{.sched2} respectively. However
3343 for @var{n} greater than nine, the output is always printed to standard
3346 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3347 same information as @option{-dRS}. For @var{n} greater than one, it
3348 also output basic block probabilities, detailed ready list information
3349 and unit/insn info. For @var{n} greater than two, it includes RTL
3350 at abort point, control-flow and regions info. And for @var{n} over
3351 four, @option{-fsched-verbose} also includes dependence info.
3355 Store the usual ``temporary'' intermediate files permanently; place them
3356 in the current directory and name them based on the source file. Thus,
3357 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3358 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3359 preprocessed @file{foo.i} output file even though the compiler now
3360 normally uses an integrated preprocessor.
3364 Report the CPU time taken by each subprocess in the compilation
3365 sequence. For C source files, this is the compiler proper and assembler
3366 (plus the linker if linking is done). The output looks like this:
3373 The first number on each line is the ``user time,'' that is time spent
3374 executing the program itself. The second number is ``system time,''
3375 time spent executing operating system routines on behalf of the program.
3376 Both numbers are in seconds.
3378 @item -print-file-name=@var{library}
3379 @opindex print-file-name
3380 Print the full absolute name of the library file @var{library} that
3381 would be used when linking---and don't do anything else. With this
3382 option, GCC does not compile or link anything; it just prints the
3385 @item -print-multi-directory
3386 @opindex print-multi-directory
3387 Print the directory name corresponding to the multilib selected by any
3388 other switches present in the command line. This directory is supposed
3389 to exist in @env{GCC_EXEC_PREFIX}.
3391 @item -print-multi-lib
3392 @opindex print-multi-lib
3393 Print the mapping from multilib directory names to compiler switches
3394 that enable them. The directory name is separated from the switches by
3395 @samp{;}, and each switch starts with an @samp{@@} instead of the
3396 @samp{-}, without spaces between multiple switches. This is supposed to
3397 ease shell-processing.
3399 @item -print-prog-name=@var{program}
3400 @opindex print-prog-name
3401 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3403 @item -print-libgcc-file-name
3404 @opindex print-libgcc-file-name
3405 Same as @option{-print-file-name=libgcc.a}.
3407 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3408 but you do want to link with @file{libgcc.a}. You can do
3411 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3414 @item -print-search-dirs
3415 @opindex print-search-dirs
3416 Print the name of the configured installation directory and a list of
3417 program and library directories gcc will search---and don't do anything else.
3419 This is useful when gcc prints the error message
3420 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3421 To resolve this you either need to put @file{cpp0} and the other compiler
3422 components where gcc expects to find them, or you can set the environment
3423 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3424 Don't forget the trailing '/'.
3425 @xref{Environment Variables}.
3428 @opindex dumpmachine
3429 Print the compiler's target machine (for example,
3430 @samp{i686-pc-linux-gnu})---and don't do anything else.
3433 @opindex dumpversion
3434 Print the compiler version (for example, @samp{3.0})---and don't do
3439 Print the compiler's built-in specs---and don't do anything else. (This
3440 is used when GCC itself is being built.) @xref{Spec Files}.
3442 @item -feliminate-unused-debug-types
3443 @opindex feliminate-unused-debug-types
3444 Normally, when producing DWARF2 output, GCC will emit debugging
3445 information for all types declared in a compilation
3446 unit, regardless of whether or not they are actually used
3447 in that compilation unit. Sometimes this is useful, such as
3448 if, in the debugger, you want to cast a value to a type that is
3449 not actually used in your program (but is declared). More often,
3450 however, this results in a significant amount of wasted space.
3451 With this option, GCC will avoid producing debug symbol output
3452 for types that are nowhere used in the source file being compiled.
3455 @node Optimize Options
3456 @section Options That Control Optimization
3457 @cindex optimize options
3458 @cindex options, optimization
3460 These options control various sorts of optimizations.
3462 Without any optimization option, the compiler's goal is to reduce the
3463 cost of compilation and to make debugging produce the expected
3464 results. Statements are independent: if you stop the program with a
3465 breakpoint between statements, you can then assign a new value to any
3466 variable or change the program counter to any other statement in the
3467 function and get exactly the results you would expect from the source
3470 Turning on optimization flags makes the compiler attempt to improve
3471 the performance and/or code size at the expense of compilation time
3472 and possibly the ability to debug the program.
3474 Not all optimizations are controlled directly by a flag. Only
3475 optimizations that have a flag are listed.
3482 Optimize. Optimizing compilation takes somewhat more time, and a lot
3483 more memory for a large function.
3485 With @option{-O}, the compiler tries to reduce code size and execution
3486 time, without performing any optimizations that take a great deal of
3489 @option{-O} turns on the following optimization flags:
3490 @gccoptlist{-fdefer-pop @gol
3491 -fmerge-constants @gol
3493 -floop-optimize @gol
3495 -fif-conversion @gol
3496 -fif-conversion2 @gol
3497 -fdelayed-branch @gol
3498 -fguess-branch-probability @gol
3501 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3502 where doing so does not interfere with debugging.
3506 Optimize even more. GCC performs nearly all supported optimizations
3507 that do not involve a space-speed tradeoff. The compiler does not
3508 perform loop unrolling or function inlining when you specify @option{-O2}.
3509 As compared to @option{-O}, this option increases both compilation time
3510 and the performance of the generated code.
3512 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3513 also turns on the following optimization flags:
3514 @gccoptlist{-fforce-mem @gol
3515 -foptimize-sibling-calls @gol
3516 -fstrength-reduce @gol
3517 -fcse-follow-jumps -fcse-skip-blocks @gol
3518 -frerun-cse-after-loop -frerun-loop-opt @gol
3519 -fgcse -fgcse-lm -fgcse-sm @gol
3520 -fdelete-null-pointer-checks @gol
3521 -fexpensive-optimizations @gol
3523 -fschedule-insns -fschedule-insns2 @gol
3524 -fsched-interblock -fsched-spec @gol
3527 -freorder-blocks -freorder-functions @gol
3528 -fstrict-aliasing @gol
3529 -falign-functions -falign-jumps @gol
3530 -falign-loops -falign-labels}
3532 Please note the warning under @option{-fgcse} about
3533 invoking @option{-O2} on programs that use computed gotos.
3537 Optimize yet more. @option{-O3} turns on all optimizations specified by
3538 @option{-O2} and also turns on the @option{-finline-functions},
3539 @option{-funit-at-a-time} and @option{-frename-registers} options.
3543 Do not optimize. This is the default.
3547 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3548 do not typically increase code size. It also performs further
3549 optimizations designed to reduce code size.
3551 @option{-Os} disables the following optimization flags:
3552 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3553 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3555 If you use multiple @option{-O} options, with or without level numbers,
3556 the last such option is the one that is effective.
3559 Options of the form @option{-f@var{flag}} specify machine-independent
3560 flags. Most flags have both positive and negative forms; the negative
3561 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3562 below, only one of the forms is listed---the one you typically will
3563 use. You can figure out the other form by either removing @samp{no-}
3566 The following options control specific optimizations. They are either
3567 activated by @option{-O} options or are related to ones that are. You
3568 can use the following flags in the rare cases when ``fine-tuning'' of
3569 optimizations to be performed is desired.
3572 @item -fno-default-inline
3573 @opindex fno-default-inline
3574 Do not make member functions inline by default merely because they are
3575 defined inside the class scope (C++ only). Otherwise, when you specify
3576 @w{@option{-O}}, member functions defined inside class scope are compiled
3577 inline by default; i.e., you don't need to add @samp{inline} in front of
3578 the member function name.
3580 @item -fno-defer-pop
3581 @opindex fno-defer-pop
3582 Always pop the arguments to each function call as soon as that function
3583 returns. For machines which must pop arguments after a function call,
3584 the compiler normally lets arguments accumulate on the stack for several
3585 function calls and pops them all at once.
3587 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3591 Force memory operands to be copied into registers before doing
3592 arithmetic on them. This produces better code by making all memory
3593 references potential common subexpressions. When they are not common
3594 subexpressions, instruction combination should eliminate the separate
3597 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3600 @opindex fforce-addr
3601 Force memory address constants to be copied into registers before
3602 doing arithmetic on them. This may produce better code just as
3603 @option{-fforce-mem} may.
3605 @item -fomit-frame-pointer
3606 @opindex fomit-frame-pointer
3607 Don't keep the frame pointer in a register for functions that
3608 don't need one. This avoids the instructions to save, set up and
3609 restore frame pointers; it also makes an extra register available
3610 in many functions. @strong{It also makes debugging impossible on
3613 On some machines, such as the VAX, this flag has no effect, because
3614 the standard calling sequence automatically handles the frame pointer
3615 and nothing is saved by pretending it doesn't exist. The
3616 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3617 whether a target machine supports this flag. @xref{Registers,,Register
3618 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3620 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3622 @item -foptimize-sibling-calls
3623 @opindex foptimize-sibling-calls
3624 Optimize sibling and tail recursive calls.
3626 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3630 Don't pay attention to the @code{inline} keyword. Normally this option
3631 is used to keep the compiler from expanding any functions inline.
3632 Note that if you are not optimizing, no functions can be expanded inline.
3634 @item -finline-functions
3635 @opindex finline-functions
3636 Integrate all simple functions into their callers. The compiler
3637 heuristically decides which functions are simple enough to be worth
3638 integrating in this way.
3640 If all calls to a given function are integrated, and the function is
3641 declared @code{static}, then the function is normally not output as
3642 assembler code in its own right.
3644 Enabled at level @option{-O3}.
3646 @item -finline-limit=@var{n}
3647 @opindex finline-limit
3648 By default, gcc limits the size of functions that can be inlined. This flag
3649 allows the control of this limit for functions that are explicitly marked as
3650 inline (i.e., marked with the inline keyword or defined within the class
3651 definition in c++). @var{n} is the size of functions that can be inlined in
3652 number of pseudo instructions (not counting parameter handling). The default
3653 value of @var{n} is 600.
3654 Increasing this value can result in more inlined code at
3655 the cost of compilation time and memory consumption. Decreasing usually makes
3656 the compilation faster and less code will be inlined (which presumably
3657 means slower programs). This option is particularly useful for programs that
3658 use inlining heavily such as those based on recursive templates with C++.
3660 Inlining is actually controlled by a number of parameters, which may be
3661 specified individually by using @option{--param @var{name}=@var{value}}.
3662 The @option{-finline-limit=@var{n}} option sets some of these parameters
3666 @item max-inline-insns
3668 @item max-inline-insns-single
3669 is set to @var{n}/2.
3670 @item max-inline-insns-auto
3671 is set to @var{n}/2.
3672 @item min-inline-insns
3673 is set to 130 or @var{n}/4, whichever is smaller.
3674 @item max-inline-insns-rtl
3678 Using @option{-finline-limit=600} thus results in the default settings
3679 for these parameters. See below for a documentation of the individual
3680 parameters controlling inlining.
3682 @emph{Note:} pseudo instruction represents, in this particular context, an
3683 abstract measurement of function's size. In no way, it represents a count
3684 of assembly instructions and as such its exact meaning might change from one
3685 release to an another.
3687 @item -fkeep-inline-functions
3688 @opindex fkeep-inline-functions
3689 Even if all calls to a given function are integrated, and the function
3690 is declared @code{static}, nevertheless output a separate run-time
3691 callable version of the function. This switch does not affect
3692 @code{extern inline} functions.
3694 @item -fkeep-static-consts
3695 @opindex fkeep-static-consts
3696 Emit variables declared @code{static const} when optimization isn't turned
3697 on, even if the variables aren't referenced.
3699 GCC enables this option by default. If you want to force the compiler to
3700 check if the variable was referenced, regardless of whether or not
3701 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3703 @item -fmerge-constants
3704 Attempt to merge identical constants (string constants and floating point
3705 constants) across compilation units.
3707 This option is the default for optimized compilation if the assembler and
3708 linker support it. Use @option{-fno-merge-constants} to inhibit this
3711 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3713 @item -fmerge-all-constants
3714 Attempt to merge identical constants and identical variables.
3716 This option implies @option{-fmerge-constants}. In addition to
3717 @option{-fmerge-constants} this considers e.g. even constant initialized
3718 arrays or initialized constant variables with integral or floating point
3719 types. Languages like C or C++ require each non-automatic variable to
3720 have distinct location, so using this option will result in non-conforming
3725 Use a graph coloring register allocator. Currently this option is meant
3726 for testing, so we are interested to hear about miscompilations with
3729 @item -fno-branch-count-reg
3730 @opindex fno-branch-count-reg
3731 Do not use ``decrement and branch'' instructions on a count register,
3732 but instead generate a sequence of instructions that decrement a
3733 register, compare it against zero, then branch based upon the result.
3734 This option is only meaningful on architectures that support such
3735 instructions, which include x86, PowerPC, IA-64 and S/390.
3737 The default is @option{-fbranch-count-reg}, enabled when
3738 @option{-fstrength-reduce} is enabled.
3740 @item -fno-function-cse
3741 @opindex fno-function-cse
3742 Do not put function addresses in registers; make each instruction that
3743 calls a constant function contain the function's address explicitly.
3745 This option results in less efficient code, but some strange hacks
3746 that alter the assembler output may be confused by the optimizations
3747 performed when this option is not used.
3749 The default is @option{-ffunction-cse}
3751 @item -fno-zero-initialized-in-bss
3752 @opindex fno-zero-initialized-in-bss
3753 If the target supports a BSS section, GCC by default puts variables that
3754 are initialized to zero into BSS@. This can save space in the resulting
3757 This option turns off this behavior because some programs explicitly
3758 rely on variables going to the data section. E.g., so that the
3759 resulting executable can find the beginning of that section and/or make
3760 assumptions based on that.
3762 The default is @option{-fzero-initialized-in-bss}.
3764 @item -fstrength-reduce
3765 @opindex fstrength-reduce
3766 Perform the optimizations of loop strength reduction and
3767 elimination of iteration variables.
3769 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3771 @item -fthread-jumps
3772 @opindex fthread-jumps
3773 Perform optimizations where we check to see if a jump branches to a
3774 location where another comparison subsumed by the first is found. If
3775 so, the first branch is redirected to either the destination of the
3776 second branch or a point immediately following it, depending on whether
3777 the condition is known to be true or false.
3779 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3781 @item -fcse-follow-jumps
3782 @opindex fcse-follow-jumps
3783 In common subexpression elimination, scan through jump instructions
3784 when the target of the jump is not reached by any other path. For
3785 example, when CSE encounters an @code{if} statement with an
3786 @code{else} clause, CSE will follow the jump when the condition
3789 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3791 @item -fcse-skip-blocks
3792 @opindex fcse-skip-blocks
3793 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3794 follow jumps which conditionally skip over blocks. When CSE
3795 encounters a simple @code{if} statement with no else clause,
3796 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3797 body of the @code{if}.
3799 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3801 @item -frerun-cse-after-loop
3802 @opindex frerun-cse-after-loop
3803 Re-run common subexpression elimination after loop optimizations has been
3806 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3808 @item -frerun-loop-opt
3809 @opindex frerun-loop-opt
3810 Run the loop optimizer twice.
3812 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3816 Perform a global common subexpression elimination pass.
3817 This pass also performs global constant and copy propagation.
3819 @emph{Note:} When compiling a program using computed gotos, a GCC
3820 extension, you may get better runtime performance if you disable
3821 the global common subexpression elimination pass by adding
3822 @option{-fno-gcse} to the command line.
3824 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3828 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3829 attempt to move loads which are only killed by stores into themselves. This
3830 allows a loop containing a load/store sequence to be changed to a load outside
3831 the loop, and a copy/store within the loop.
3833 Enabled by default when gcse is enabled.
3837 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3838 subexpression elimination. This pass will attempt to move stores out of loops.
3839 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3840 can be changed to a load before the loop and a store after the loop.
3842 Enabled by default when gcse is enabled.
3844 @item -floop-optimize
3845 @opindex floop-optimize
3846 Perform loop optimizations: move constant expressions out of loops, simplify
3847 exit test conditions and optionally do strength-reduction and loop unrolling as
3850 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3852 @item -fcrossjumping
3853 @opindex crossjumping
3854 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3855 resulting code may or may not perform better than without cross-jumping.
3857 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3859 @item -fif-conversion
3860 @opindex if-conversion
3861 Attempt to transform conditional jumps into branch-less equivalents. This
3862 include use of conditional moves, min, max, set flags and abs instructions, and
3863 some tricks doable by standard arithmetics. The use of conditional execution
3864 on chips where it is available is controlled by @code{if-conversion2}.
3866 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3868 @item -fif-conversion2
3869 @opindex if-conversion2
3870 Use conditional execution (where available) to transform conditional jumps into
3871 branch-less equivalents.
3873 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3875 @item -fdelete-null-pointer-checks
3876 @opindex fdelete-null-pointer-checks
3877 Use global dataflow analysis to identify and eliminate useless checks
3878 for null pointers. The compiler assumes that dereferencing a null
3879 pointer would have halted the program. If a pointer is checked after
3880 it has already been dereferenced, it cannot be null.
3882 In some environments, this assumption is not true, and programs can
3883 safely dereference null pointers. Use
3884 @option{-fno-delete-null-pointer-checks} to disable this optimization
3885 for programs which depend on that behavior.
3887 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3889 @item -fexpensive-optimizations
3890 @opindex fexpensive-optimizations
3891 Perform a number of minor optimizations that are relatively expensive.
3893 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3895 @item -foptimize-register-move
3897 @opindex foptimize-register-move
3899 Attempt to reassign register numbers in move instructions and as
3900 operands of other simple instructions in order to maximize the amount of
3901 register tying. This is especially helpful on machines with two-operand
3904 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3907 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3909 @item -fdelayed-branch
3910 @opindex fdelayed-branch
3911 If supported for the target machine, attempt to reorder instructions
3912 to exploit instruction slots available after delayed branch
3915 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3917 @item -fschedule-insns
3918 @opindex fschedule-insns
3919 If supported for the target machine, attempt to reorder instructions to
3920 eliminate execution stalls due to required data being unavailable. This
3921 helps machines that have slow floating point or memory load instructions
3922 by allowing other instructions to be issued until the result of the load
3923 or floating point instruction is required.
3925 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3927 @item -fschedule-insns2
3928 @opindex fschedule-insns2
3929 Similar to @option{-fschedule-insns}, but requests an additional pass of
3930 instruction scheduling after register allocation has been done. This is
3931 especially useful on machines with a relatively small number of
3932 registers and where memory load instructions take more than one cycle.
3934 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3936 @item -fno-sched-interblock
3937 @opindex fno-sched-interblock
3938 Don't schedule instructions across basic blocks. This is normally
3939 enabled by default when scheduling before register allocation, i.e.@:
3940 with @option{-fschedule-insns} or at @option{-O2} or higher.
3942 @item -fno-sched-spec
3943 @opindex fno-sched-spec
3944 Don't allow speculative motion of non-load instructions. This is normally
3945 enabled by default when scheduling before register allocation, i.e.@:
3946 with @option{-fschedule-insns} or at @option{-O2} or higher.
3948 @item -fsched-spec-load
3949 @opindex fsched-spec-load
3950 Allow speculative motion of some load instructions. This only makes
3951 sense when scheduling before register allocation, i.e.@: with
3952 @option{-fschedule-insns} or at @option{-O2} or higher.
3954 @item -fsched-spec-load-dangerous
3955 @opindex fsched-spec-load-dangerous
3956 Allow speculative motion of more load instructions. This only makes
3957 sense when scheduling before register allocation, i.e.@: with
3958 @option{-fschedule-insns} or at @option{-O2} or higher.
3960 @item -fsched2-use-superblocks
3961 @opindex fsched2-use-superblocks
3962 When scheduling after register allocation, do use superblock scheduling
3963 algorithm. Superblock scheduling allows motion across basic block boundaries
3964 resulting on faster schedules. This option is experimental, as not all machine
3965 descriptions used by GCC model the CPU closely enough to avoid unreliable
3966 results from the algorithm.
3968 This only makes sense when scheduling after register allocation, i.e.@: with
3969 @option{-fschedule-insns2} or at @option{-O2} or higher.
3971 @item -fsched2-use-traces
3972 @opindex fsched2-use-traces
3973 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
3974 allocation and additionally perform code duplication in order to increase the
3975 size of superblocks using tracer pass. See @option{-ftracer} for details on
3978 This mode should produce faster but significantly longer programs. Also
3979 without @code{-fbranch-probabilities} the traces constructed may not match the
3980 reality and hurt the performance. This only makes
3981 sense when scheduling after register allocation, i.e.@: with
3982 @option{-fschedule-insns2} or at @option{-O2} or higher.
3984 @item -fcaller-saves
3985 @opindex fcaller-saves
3986 Enable values to be allocated in registers that will be clobbered by
3987 function calls, by emitting extra instructions to save and restore the
3988 registers around such calls. Such allocation is done only when it
3989 seems to result in better code than would otherwise be produced.
3991 This option is always enabled by default on certain machines, usually
3992 those which have no call-preserved registers to use instead.
3994 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3996 @item -fmove-all-movables
3997 @opindex fmove-all-movables
3998 Forces all invariant computations in loops to be moved
4001 @item -freduce-all-givs
4002 @opindex freduce-all-givs
4003 Forces all general-induction variables in loops to be
4006 @emph{Note:} When compiling programs written in Fortran,
4007 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4008 by default when you use the optimizer.
4010 These options may generate better or worse code; results are highly
4011 dependent on the structure of loops within the source code.
4013 These two options are intended to be removed someday, once
4014 they have helped determine the efficacy of various
4015 approaches to improving loop optimizations.
4017 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4018 know how use of these options affects
4019 the performance of your production code.
4020 We're very interested in code that runs @emph{slower}
4021 when these options are @emph{enabled}.
4024 @itemx -fno-peephole2
4025 @opindex fno-peephole
4026 @opindex fno-peephole2
4027 Disable any machine-specific peephole optimizations. The difference
4028 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4029 are implemented in the compiler; some targets use one, some use the
4030 other, a few use both.
4032 @option{-fpeephole} is enabled by default.
4033 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4035 @item -fno-guess-branch-probability
4036 @opindex fno-guess-branch-probability
4037 Do not guess branch probabilities using a randomized model.
4039 Sometimes gcc will opt to use a randomized model to guess branch
4040 probabilities, when none are available from either profiling feedback
4041 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4042 different runs of the compiler on the same program may produce different
4045 In a hard real-time system, people don't want different runs of the
4046 compiler to produce code that has different behavior; minimizing
4047 non-determinism is of paramount import. This switch allows users to
4048 reduce non-determinism, possibly at the expense of inferior
4051 The default is @option{-fguess-branch-probability} at levels
4052 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4054 @item -freorder-blocks
4055 @opindex freorder-blocks
4056 Reorder basic blocks in the compiled function in order to reduce number of
4057 taken branches and improve code locality.
4059 Enabled at levels @option{-O2}, @option{-O3}.
4061 @item -freorder-functions
4062 @opindex freorder-functions
4063 Reorder basic blocks in the compiled function in order to reduce number of
4064 taken branches and improve code locality. This is implemented by using special
4065 subsections @code{text.hot} for most frequently executed functions and
4066 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4067 the linker so object file format must support named sections and linker must
4068 place them in a reasonable way.
4070 Also profile feedback must be available in to make this option effective. See
4071 @option{-fprofile-arcs} for details.
4073 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4075 @item -fstrict-aliasing
4076 @opindex fstrict-aliasing
4077 Allows the compiler to assume the strictest aliasing rules applicable to
4078 the language being compiled. For C (and C++), this activates
4079 optimizations based on the type of expressions. In particular, an
4080 object of one type is assumed never to reside at the same address as an
4081 object of a different type, unless the types are almost the same. For
4082 example, an @code{unsigned int} can alias an @code{int}, but not a
4083 @code{void*} or a @code{double}. A character type may alias any other
4086 Pay special attention to code like this:
4099 The practice of reading from a different union member than the one most
4100 recently written to (called ``type-punning'') is common. Even with
4101 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4102 is accessed through the union type. So, the code above will work as
4103 expected. However, this code might not:
4114 Every language that wishes to perform language-specific alias analysis
4115 should define a function that computes, given an @code{tree}
4116 node, an alias set for the node. Nodes in different alias sets are not
4117 allowed to alias. For an example, see the C front-end function
4118 @code{c_get_alias_set}.
4120 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4122 @item -falign-functions
4123 @itemx -falign-functions=@var{n}
4124 @opindex falign-functions
4125 Align the start of functions to the next power-of-two greater than
4126 @var{n}, skipping up to @var{n} bytes. For instance,
4127 @option{-falign-functions=32} aligns functions to the next 32-byte
4128 boundary, but @option{-falign-functions=24} would align to the next
4129 32-byte boundary only if this can be done by skipping 23 bytes or less.
4131 @option{-fno-align-functions} and @option{-falign-functions=1} are
4132 equivalent and mean that functions will not be aligned.
4134 Some assemblers only support this flag when @var{n} is a power of two;
4135 in that case, it is rounded up.
4137 If @var{n} is not specified or is zero, use a machine-dependent default.
4139 Enabled at levels @option{-O2}, @option{-O3}.
4141 @item -falign-labels
4142 @itemx -falign-labels=@var{n}
4143 @opindex falign-labels
4144 Align all branch targets to a power-of-two boundary, skipping up to
4145 @var{n} bytes like @option{-falign-functions}. This option can easily
4146 make code slower, because it must insert dummy operations for when the
4147 branch target is reached in the usual flow of the code.
4149 @option{-fno-align-labels} and @option{-falign-labels=1} are
4150 equivalent and mean that labels will not be aligned.
4152 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4153 are greater than this value, then their values are used instead.
4155 If @var{n} is not specified or is zero, use a machine-dependent default
4156 which is very likely to be @samp{1}, meaning no alignment.
4158 Enabled at levels @option{-O2}, @option{-O3}.
4161 @itemx -falign-loops=@var{n}
4162 @opindex falign-loops
4163 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4164 like @option{-falign-functions}. The hope is that the loop will be
4165 executed many times, which will make up for any execution of the dummy
4168 @option{-fno-align-loops} and @option{-falign-loops=1} are
4169 equivalent and mean that loops will not be aligned.
4171 If @var{n} is not specified or is zero, use a machine-dependent default.
4173 Enabled at levels @option{-O2}, @option{-O3}.
4176 @itemx -falign-jumps=@var{n}
4177 @opindex falign-jumps
4178 Align branch targets to a power-of-two boundary, for branch targets
4179 where the targets can only be reached by jumping, skipping up to @var{n}
4180 bytes like @option{-falign-functions}. In this case, no dummy operations
4183 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4184 equivalent and mean that loops will not be aligned.
4186 If @var{n} is not specified or is zero, use a machine-dependent default.
4188 Enabled at levels @option{-O2}, @option{-O3}.
4190 @item -frename-registers
4191 @opindex frename-registers
4192 Attempt to avoid false dependencies in scheduled code by making use
4193 of registers left over after register allocation. This optimization
4194 will most benefit processors with lots of registers. It can, however,
4195 make debugging impossible, since variables will no longer stay in
4196 a ``home register''.
4198 Enabled at levels @option{-O3}.
4200 @item -fno-cprop-registers
4201 @opindex fno-cprop-registers
4202 After register allocation and post-register allocation instruction splitting,
4203 we perform a copy-propagation pass to try to reduce scheduling dependencies
4204 and occasionally eliminate the copy.
4206 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4210 The following options control compiler behavior regarding floating
4211 point arithmetic. These options trade off between speed and
4212 correctness. All must be specifically enabled.
4216 @opindex ffloat-store
4217 Do not store floating point variables in registers, and inhibit other
4218 options that might change whether a floating point value is taken from a
4221 @cindex floating point precision
4222 This option prevents undesirable excess precision on machines such as
4223 the 68000 where the floating registers (of the 68881) keep more
4224 precision than a @code{double} is supposed to have. Similarly for the
4225 x86 architecture. For most programs, the excess precision does only
4226 good, but a few programs rely on the precise definition of IEEE floating
4227 point. Use @option{-ffloat-store} for such programs, after modifying
4228 them to store all pertinent intermediate computations into variables.
4232 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4233 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
4234 @option{-fno-signaling-nans}.
4236 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4238 This option should never be turned on by any @option{-O} option since
4239 it can result in incorrect output for programs which depend on
4240 an exact implementation of IEEE or ISO rules/specifications for
4243 @item -fno-math-errno
4244 @opindex fno-math-errno
4245 Do not set ERRNO after calling math functions that are executed
4246 with a single instruction, e.g., sqrt. A program that relies on
4247 IEEE exceptions for math error handling may want to use this flag
4248 for speed while maintaining IEEE arithmetic compatibility.
4250 This option should never be turned on by any @option{-O} option since
4251 it can result in incorrect output for programs which depend on
4252 an exact implementation of IEEE or ISO rules/specifications for
4255 The default is @option{-fmath-errno}.
4257 @item -funsafe-math-optimizations
4258 @opindex funsafe-math-optimizations
4259 Allow optimizations for floating-point arithmetic that (a) assume
4260 that arguments and results are valid and (b) may violate IEEE or
4261 ANSI standards. When used at link-time, it may include libraries
4262 or startup files that change the default FPU control word or other
4263 similar optimizations.
4265 This option should never be turned on by any @option{-O} option since
4266 it can result in incorrect output for programs which depend on
4267 an exact implementation of IEEE or ISO rules/specifications for
4270 The default is @option{-fno-unsafe-math-optimizations}.
4272 @item -ffinite-math-only
4273 @opindex ffinite-math-only
4274 Allow optimizations for floating-point arithmetic that assume
4275 that arguments and results are not NaNs or +-Infs.
4277 This option should never be turned on by any @option{-O} option since
4278 it can result in incorrect output for programs which depend on
4279 an exact implementation of IEEE or ISO rules/specifications.
4281 The default is @option{-fno-finite-math-only}.
4283 @item -fno-trapping-math
4284 @opindex fno-trapping-math
4285 Compile code assuming that floating-point operations cannot generate
4286 user-visible traps. These traps include division by zero, overflow,
4287 underflow, inexact result and invalid operation. This option implies
4288 @option{-fno-signaling-nans}. Setting this option may allow faster
4289 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4291 This option should never be turned on by any @option{-O} option since
4292 it can result in incorrect output for programs which depend on
4293 an exact implementation of IEEE or ISO rules/specifications for
4296 The default is @option{-ftrapping-math}.
4298 @item -fsignaling-nans
4299 @opindex fsignaling-nans
4300 Compile code assuming that IEEE signaling NaNs may generate user-visible
4301 traps during floating-point operations. Setting this option disables
4302 optimizations that may change the number of exceptions visible with
4303 signaling NaNs. This option implies @option{-ftrapping-math}.
4305 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4308 The default is @option{-fno-signaling-nans}.
4310 This option is experimental and does not currently guarantee to
4311 disable all GCC optimizations that affect signaling NaN behavior.
4313 @item -fsingle-precision-constant
4314 @opindex fsingle-precision-constant
4315 Treat floating point constant as single precision constant instead of
4316 implicitly converting it to double precision constant.
4321 The following options control optimizations that may improve
4322 performance, but are not enabled by any @option{-O} options. This
4323 section includes experimental options that may produce broken code.
4326 @item -fbranch-probabilities
4327 @opindex fbranch-probabilities
4328 After running a program compiled with @option{-fprofile-arcs}
4329 (@pxref{Debugging Options,, Options for Debugging Your Program or
4330 @command{gcc}}), you can compile it a second time using
4331 @option{-fbranch-probabilities}, to improve optimizations based on
4332 the number of times each branch was taken. When the program
4333 compiled with @option{-fprofile-arcs} exits it saves arc execution
4334 counts to a file called @file{@var{sourcename}.da} for each source
4335 file The information in this data file is very dependent on the
4336 structure of the generated code, so you must use the same source code
4337 and the same optimization options for both compilations.
4339 With @option{-fbranch-probabilities}, GCC puts a
4340 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4341 These can be used to improve optimization. Currently, they are only
4342 used in one place: in @file{reorg.c}, instead of guessing which path a
4343 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4344 exactly determine which path is taken more often.
4346 @item -fprofile-values
4347 @opindex fprofile-values
4348 If combined with @option{-fprofile-arcs}, it adds code so that some
4349 data about values of expressions in the program is gathered.
4353 Use a graph coloring register allocator. Currently this option is meant
4354 for testing, so we are interested to hear about miscompilations with
4359 Perform tail duplication to enlarge superblock size. This transformation
4360 simplifies the control flow of the function allowing other optimizations to do
4363 @item -funit-at-a-time
4364 @opindex funit-at-a-time
4365 Parse the whole compilation unit before starting to produce code. This allows some
4366 extra optimizations to take place but consumes more memory.
4368 @item -funroll-loops
4369 @opindex funroll-loops
4370 Unroll loops whose number of iterations can be determined at compile time or
4371 upon entry to the loop. @option{-funroll-loops} implies
4372 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4373 (i.e. complete removal of loops with small constant number of iterations).
4374 This option makes code larger, and may or may not make it run faster.
4376 @item -funroll-all-loops
4377 @opindex funroll-all-loops
4378 Unroll all loops, even if their number of iterations is uncertain when
4379 the loop is entered. This usually makes programs run more slowly.
4380 @option{-funroll-all-loops} implies the same options as
4381 @option{-funroll-loops}.
4384 @opindex fpeel-loops
4385 Peels the loops for that there is enough information that they do not
4386 roll much (from profile feedback). It also turns on complete loop peeling
4387 (i.e. complete removal of loops with small constant number of iterations).
4389 @item -funswitch-loops
4390 @opindex funswitch-loops
4391 Move branches with loop invariant conditions out of the loop, with duplicates
4392 of the loop on both branches (modified according to result of the condition).
4394 @item -fold-unroll-loops
4395 @opindex fold-unroll-loops
4396 Unroll loops whose number of iterations can be determined at compile
4397 time or upon entry to the loop, using the old loop unroller whose loop
4398 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4399 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4400 option makes code larger, and may or may not make it run faster.
4402 @item -fold-unroll-all-loops
4403 @opindex fold-unroll-all-loops
4404 Unroll all loops, even if their number of iterations is uncertain when
4405 the loop is entered. This is done using the old loop unroller whose loop
4406 recognition is based on notes from frontend. This usually makes programs run more slowly.
4407 @option{-fold-unroll-all-loops} implies the same options as
4408 @option{-fold-unroll-loops}.
4410 @item -funswitch-loops
4411 @opindex funswitch-loops
4412 Move branches with loop invariant conditions out of the loop, with duplicates
4413 of the loop on both branches (modified according to result of the condition).
4415 @item -funswitch-loops
4416 @opindex funswitch-loops
4417 Move branches with loop invariant conditions out of the loop, with duplicates
4418 of the loop on both branches (modified according to result of the condition).
4420 @item -fprefetch-loop-arrays
4421 @opindex fprefetch-loop-arrays
4422 If supported by the target machine, generate instructions to prefetch
4423 memory to improve the performance of loops that access large arrays.
4425 Disabled at level @option{-Os}.
4427 @item -ffunction-sections
4428 @itemx -fdata-sections
4429 @opindex ffunction-sections
4430 @opindex fdata-sections
4431 Place each function or data item into its own section in the output
4432 file if the target supports arbitrary sections. The name of the
4433 function or the name of the data item determines the section's name
4436 Use these options on systems where the linker can perform optimizations
4437 to improve locality of reference in the instruction space. Most systems
4438 using the ELF object format and SPARC processors running Solaris 2 have
4439 linkers with such optimizations. AIX may have these optimizations in
4442 Only use these options when there are significant benefits from doing
4443 so. When you specify these options, the assembler and linker will
4444 create larger object and executable files and will also be slower.
4445 You will not be able to use @code{gprof} on all systems if you
4446 specify this option and you may have problems with debugging if
4447 you specify both this option and @option{-g}.
4451 Perform optimizations in static single assignment form. Each function's
4452 flow graph is translated into SSA form, optimizations are performed, and
4453 the flow graph is translated back from SSA form. Users should not
4454 specify this option, since it is not yet ready for production use.
4458 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4459 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4463 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4464 Like @option{-fssa}, this is an experimental feature.
4466 @item -fbranch-target-load-optimize
4467 @opindex fbranch-target-load-optimize
4468 Perform branch target register load optimization before prologue / epilogue
4470 The use of target registers can typically be exposed only during reload,
4471 thus hoisting loads out of loops and doing inter-block scheduling needs
4472 a separate optimization pass.
4474 @item -fbranch-target-load-optimize2
4475 @opindex fbranch-target-load-optimize2
4476 Perform branch target register load optimization after prologue / epilogue
4482 @item --param @var{name}=@var{value}
4484 In some places, GCC uses various constants to control the amount of
4485 optimization that is done. For example, GCC will not inline functions
4486 that contain more that a certain number of instructions. You can
4487 control some of these constants on the command-line using the
4488 @option{--param} option.
4490 In each case, the @var{value} is an integer. The allowable choices for
4491 @var{name} are given in the following table:
4494 @item max-crossjump-edges
4495 The maximum number of incoming edges to consider for crossjumping.
4496 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4497 the number of edges incoming to each block. Increasing values mean
4498 more aggressive optimization, making the compile time increase with
4499 probably small improvement in executable size.
4501 @item max-delay-slot-insn-search
4502 The maximum number of instructions to consider when looking for an
4503 instruction to fill a delay slot. If more than this arbitrary number of
4504 instructions is searched, the time savings from filling the delay slot
4505 will be minimal so stop searching. Increasing values mean more
4506 aggressive optimization, making the compile time increase with probably
4507 small improvement in executable run time.
4509 @item max-delay-slot-live-search
4510 When trying to fill delay slots, the maximum number of instructions to
4511 consider when searching for a block with valid live register
4512 information. Increasing this arbitrarily chosen value means more
4513 aggressive optimization, increasing the compile time. This parameter
4514 should be removed when the delay slot code is rewritten to maintain the
4517 @item max-gcse-memory
4518 The approximate maximum amount of memory that will be allocated in
4519 order to perform the global common subexpression elimination
4520 optimization. If more memory than specified is required, the
4521 optimization will not be done.
4523 @item max-gcse-passes
4524 The maximum number of passes of GCSE to run.
4526 @item max-pending-list-length
4527 The maximum number of pending dependencies scheduling will allow
4528 before flushing the current state and starting over. Large functions
4529 with few branches or calls can create excessively large lists which
4530 needlessly consume memory and resources.
4532 @item max-inline-insns-single
4533 Several parameters control the tree inliner used in gcc.
4534 This number sets the maximum number of instructions (counted in gcc's
4535 internal representation) in a single function that the tree inliner
4536 will consider for inlining. This only affects functions declared
4537 inline and methods implemented in a class declaration (C++).
4538 The default value is 100.
4540 @item max-inline-insns-auto
4541 When you use @option{-finline-functions} (included in @option{-O3}),
4542 a lot of functions that would otherwise not be considered for inlining
4543 by the compiler will be investigated. To those functions, a different
4544 (more restrictive) limit compared to functions declared inline can
4546 The default value is 100.
4548 @item max-inline-insns
4549 The tree inliner does decrease the allowable size for single functions
4550 to be inlined after we already inlined the number of instructions
4551 given here by repeated inlining. This number should be a factor of
4552 two or more larger than the single function limit.
4553 Higher numbers result in better runtime performance, but incur higher
4554 compile-time resource (CPU time, memory) requirements and result in
4555 larger binaries. Very high values are not advisable, as too large
4556 binaries may adversely affect runtime performance.
4557 The default value is 200.
4559 @item max-inline-slope
4560 After exceeding the maximum number of inlined instructions by repeated
4561 inlining, a linear function is used to decrease the allowable size
4562 for single functions. The slope of that function is the negative
4563 reciprocal of the number specified here.
4564 The default value is 32.
4566 @item min-inline-insns
4567 The repeated inlining is throttled more and more by the linear function
4568 after exceeding the limit. To avoid too much throttling, a minimum for
4569 this function is specified here to allow repeated inlining for very small
4570 functions even when a lot of repeated inlining already has been done.
4571 The default value is 10.
4573 @item max-inline-insns-rtl
4574 For languages that use the RTL inliner (this happens at a later stage
4575 than tree inlining), you can set the maximum allowable size (counted
4576 in RTL instructions) for the RTL inliner with this parameter.
4577 The default value is 600.
4580 @item max-unrolled-insns
4581 The maximum number of instructions that a loop should have if that loop
4582 is unrolled, and if the loop is unrolled, it determines how many times
4583 the loop code is unrolled.
4585 @item max-average-unrolled-insns
4586 The maximum number of instructions biased by probabilities of their execution
4587 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4588 it determines how many times the loop code is unrolled.
4590 @item max-unroll-times
4591 The maximum number of unrollings of a single loop.
4593 @item max-peeled-insns
4594 The maximum number of instructions that a loop should have if that loop
4595 is peeled, and if the loop is peeled, it determines how many times
4596 the loop code is peeled.
4598 @item max-peel-times
4599 The maximum number of peelings of a single loop.
4601 @item max-completely-peeled-insns
4602 The maximum number of insns of a completely peeled loop.
4604 @item max-completely-peel-times
4605 The maximum number of iterations of a loop to be suitable for complete peeling.
4607 @item max-unswitch-insns
4608 The maximum number of insns of an unswitched loop.
4610 @item max-unswitch-level
4611 The maximum number of branches unswitched in a single loop.
4613 @item hot-bb-count-fraction
4614 Select fraction of the maximal count of repetitions of basic block in program
4615 given basic block needs to have to be considered hot.
4617 @item hot-bb-frequency-fraction
4618 Select fraction of the maximal frequency of executions of basic block in
4619 function given basic block needs to have to be considered hot
4621 @item tracer-dynamic-coverage
4622 @itemx tracer-dynamic-coverage-feedback
4624 This value is used to limit superblock formation once the given percentage of
4625 executed instructions is covered. This limits unnecessary code size
4628 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4629 feedback is available. The real profiles (as opposed to statically estimated
4630 ones) are much less balanced allowing the threshold to be larger value.
4632 @item tracer-max-code-growth
4633 Stop tail duplication once code growth has reached given percentage. This is
4634 rather hokey argument, as most of the duplicates will be eliminated later in
4635 cross jumping, so it may be set to much higher values than is the desired code
4638 @item tracer-min-branch-ratio
4640 Stop reverse growth when the reverse probability of best edge is less than this
4641 threshold (in percent).
4643 @item tracer-min-branch-ratio
4644 @itemx tracer-min-branch-ratio-feedback
4646 Stop forward growth if the best edge do have probability lower than this
4649 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4650 compilation for profile feedback and one for compilation without. The value
4651 for compilation with profile feedback needs to be more conservative (higher) in
4652 order to make tracer effective.
4654 @item max-cse-path-length
4656 Maximum number of basic blocks on path that cse considers.
4658 @item ggc-min-expand
4660 GCC uses a garbage collector to manage its own memory allocation. This
4661 parameter specifies the minimum percentage by which the garbage
4662 collector's heap should be allowed to expand between collections.
4663 Tuning this may improve compilation speed; it has no effect on code
4666 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4667 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4668 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4669 GCC is not able to calculate RAM on a particular platform, the lower
4670 bound of 30% is used. Setting this parameter and
4671 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4672 every opportunity. This is extremely slow, but can be useful for
4675 @item ggc-min-heapsize
4677 Minimum size of the garbage collector's heap before it begins bothering
4678 to collect garbage. The first collection occurs after the heap expands
4679 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4680 tuning this may improve compilation speed, and has no effect on code
4683 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4684 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4685 available, the notion of "RAM" is the smallest of actual RAM,
4686 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4687 RAM on a particular platform, the lower bound is used. Setting this
4688 parameter very large effectively disables garbage collection. Setting
4689 this parameter and @option{ggc-min-expand} to zero causes a full
4690 collection to occur at every opportunity.
4692 @item reorder-blocks-duplicate
4693 @itemx reorder-blocks-duplicate-feedback
4695 Used by basic block reordering pass to decide whether to use unconditional
4696 branch or duplicate the code on it's destination. Code is duplicated when it's
4697 estimated size is smaller than this value multiplied by the estimated size of
4698 unconditional jump in the hot spots of the program.
4700 The @option{reorder-block-duplicate-feedback} is used only when profile
4701 feedback is available and may be set to higher values than
4702 @option{reorder-block-duplicate} since information about the hot spots is more
4707 @node Preprocessor Options
4708 @section Options Controlling the Preprocessor
4709 @cindex preprocessor options
4710 @cindex options, preprocessor
4712 These options control the C preprocessor, which is run on each C source
4713 file before actual compilation.
4715 If you use the @option{-E} option, nothing is done except preprocessing.
4716 Some of these options make sense only together with @option{-E} because
4717 they cause the preprocessor output to be unsuitable for actual
4722 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4723 and pass @var{option} directly through to the preprocessor. If
4724 @var{option} contains commas, it is split into multiple options at the
4725 commas. However, many options are modified, translated or interpreted
4726 by the compiler driver before being passed to the preprocessor, and
4727 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4728 interface is undocumented and subject to change, so whenever possible
4729 you should avoid using @option{-Wp} and let the driver handle the
4732 @item -Xpreprocessor @var{option}
4733 @opindex preprocessor
4734 Pass @var{option} as an option to the preprocessor. You can use this to
4735 supply system-specific preprocessor options which GCC does not know how to
4738 If you want to pass an option that takes an argument, you must use
4739 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4742 @include cppopts.texi
4744 @node Assembler Options
4745 @section Passing Options to the Assembler
4747 @c prevent bad page break with this line
4748 You can pass options to the assembler.
4751 @item -Wa,@var{option}
4753 Pass @var{option} as an option to the assembler. If @var{option}
4754 contains commas, it is split into multiple options at the commas.
4756 @item -Xassembler @var{option}
4758 Pass @var{option} as an option to the assembler. You can use this to
4759 supply system-specific assembler options which GCC does not know how to
4762 If you want to pass an option that takes an argument, you must use
4763 @option{-Xassembler} twice, once for the option and once for the argument.
4768 @section Options for Linking
4769 @cindex link options
4770 @cindex options, linking
4772 These options come into play when the compiler links object files into
4773 an executable output file. They are meaningless if the compiler is
4774 not doing a link step.
4778 @item @var{object-file-name}
4779 A file name that does not end in a special recognized suffix is
4780 considered to name an object file or library. (Object files are
4781 distinguished from libraries by the linker according to the file
4782 contents.) If linking is done, these object files are used as input
4791 If any of these options is used, then the linker is not run, and
4792 object file names should not be used as arguments. @xref{Overall
4796 @item -l@var{library}
4797 @itemx -l @var{library}
4799 Search the library named @var{library} when linking. (The second
4800 alternative with the library as a separate argument is only for
4801 POSIX compliance and is not recommended.)
4803 It makes a difference where in the command you write this option; the
4804 linker searches and processes libraries and object files in the order they
4805 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4806 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4807 to functions in @samp{z}, those functions may not be loaded.
4809 The linker searches a standard list of directories for the library,
4810 which is actually a file named @file{lib@var{library}.a}. The linker
4811 then uses this file as if it had been specified precisely by name.
4813 The directories searched include several standard system directories
4814 plus any that you specify with @option{-L}.
4816 Normally the files found this way are library files---archive files
4817 whose members are object files. The linker handles an archive file by
4818 scanning through it for members which define symbols that have so far
4819 been referenced but not defined. But if the file that is found is an
4820 ordinary object file, it is linked in the usual fashion. The only
4821 difference between using an @option{-l} option and specifying a file name
4822 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4823 and searches several directories.
4827 You need this special case of the @option{-l} option in order to
4828 link an Objective-C program.
4831 @opindex nostartfiles
4832 Do not use the standard system startup files when linking.
4833 The standard system libraries are used normally, unless @option{-nostdlib}
4834 or @option{-nodefaultlibs} is used.
4836 @item -nodefaultlibs
4837 @opindex nodefaultlibs
4838 Do not use the standard system libraries when linking.
4839 Only the libraries you specify will be passed to the linker.
4840 The standard startup files are used normally, unless @option{-nostartfiles}
4841 is used. The compiler may generate calls to memcmp, memset, and memcpy
4842 for System V (and ISO C) environments or to bcopy and bzero for
4843 BSD environments. These entries are usually resolved by entries in
4844 libc. These entry points should be supplied through some other
4845 mechanism when this option is specified.
4849 Do not use the standard system startup files or libraries when linking.
4850 No startup files and only the libraries you specify will be passed to
4851 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4852 for System V (and ISO C) environments or to bcopy and bzero for
4853 BSD environments. These entries are usually resolved by entries in
4854 libc. These entry points should be supplied through some other
4855 mechanism when this option is specified.
4857 @cindex @option{-lgcc}, use with @option{-nostdlib}
4858 @cindex @option{-nostdlib} and unresolved references
4859 @cindex unresolved references and @option{-nostdlib}
4860 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4861 @cindex @option{-nodefaultlibs} and unresolved references
4862 @cindex unresolved references and @option{-nodefaultlibs}
4863 One of the standard libraries bypassed by @option{-nostdlib} and
4864 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4865 that GCC uses to overcome shortcomings of particular machines, or special
4866 needs for some languages.
4867 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4868 Collection (GCC) Internals},
4869 for more discussion of @file{libgcc.a}.)
4870 In most cases, you need @file{libgcc.a} even when you want to avoid
4871 other standard libraries. In other words, when you specify @option{-nostdlib}
4872 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4873 This ensures that you have no unresolved references to internal GCC
4874 library subroutines. (For example, @samp{__main}, used to ensure C++
4875 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4876 GNU Compiler Collection (GCC) Internals}.)
4880 Produce a position independent executable on targets which support it.
4881 For predictable results, you must also specify the same set of options
4882 that were used to generate code (@option{-fpie}, @option{-fPIE},
4883 or model suboptions) when you specify this option.
4887 Remove all symbol table and relocation information from the executable.
4891 On systems that support dynamic linking, this prevents linking with the shared
4892 libraries. On other systems, this option has no effect.
4896 Produce a shared object which can then be linked with other objects to
4897 form an executable. Not all systems support this option. For predictable
4898 results, you must also specify the same set of options that were used to
4899 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4900 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4901 needs to build supplementary stub code for constructors to work. On
4902 multi-libbed systems, @samp{gcc -shared} must select the correct support
4903 libraries to link against. Failing to supply the correct flags may lead
4904 to subtle defects. Supplying them in cases where they are not necessary
4907 @item -shared-libgcc
4908 @itemx -static-libgcc
4909 @opindex shared-libgcc
4910 @opindex static-libgcc
4911 On systems that provide @file{libgcc} as a shared library, these options
4912 force the use of either the shared or static version respectively.
4913 If no shared version of @file{libgcc} was built when the compiler was
4914 configured, these options have no effect.
4916 There are several situations in which an application should use the
4917 shared @file{libgcc} instead of the static version. The most common
4918 of these is when the application wishes to throw and catch exceptions
4919 across different shared libraries. In that case, each of the libraries
4920 as well as the application itself should use the shared @file{libgcc}.
4922 Therefore, the G++ and GCJ drivers automatically add
4923 @option{-shared-libgcc} whenever you build a shared library or a main
4924 executable, because C++ and Java programs typically use exceptions, so
4925 this is the right thing to do.
4927 If, instead, you use the GCC driver to create shared libraries, you may
4928 find that they will not always be linked with the shared @file{libgcc}.
4929 If GCC finds, at its configuration time, that you have a GNU linker that
4930 does not support option @option{--eh-frame-hdr}, it will link the shared
4931 version of @file{libgcc} into shared libraries by default. Otherwise,
4932 it will take advantage of the linker and optimize away the linking with
4933 the shared version of @file{libgcc}, linking with the static version of
4934 libgcc by default. This allows exceptions to propagate through such
4935 shared libraries, without incurring relocation costs at library load
4938 However, if a library or main executable is supposed to throw or catch
4939 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4940 for the languages used in the program, or using the option
4941 @option{-shared-libgcc}, such that it is linked with the shared
4946 Bind references to global symbols when building a shared object. Warn
4947 about any unresolved references (unless overridden by the link editor
4948 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4951 @item -Xlinker @var{option}
4953 Pass @var{option} as an option to the linker. You can use this to
4954 supply system-specific linker options which GCC does not know how to
4957 If you want to pass an option that takes an argument, you must use
4958 @option{-Xlinker} twice, once for the option and once for the argument.
4959 For example, to pass @option{-assert definitions}, you must write
4960 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4961 @option{-Xlinker "-assert definitions"}, because this passes the entire
4962 string as a single argument, which is not what the linker expects.
4964 @item -Wl,@var{option}
4966 Pass @var{option} as an option to the linker. If @var{option} contains
4967 commas, it is split into multiple options at the commas.
4969 @item -u @var{symbol}
4971 Pretend the symbol @var{symbol} is undefined, to force linking of
4972 library modules to define it. You can use @option{-u} multiple times with
4973 different symbols to force loading of additional library modules.
4976 @node Directory Options
4977 @section Options for Directory Search
4978 @cindex directory options
4979 @cindex options, directory search
4982 These options specify directories to search for header files, for
4983 libraries and for parts of the compiler:
4988 Add the directory @var{dir} to the head of the list of directories to be
4989 searched for header files. This can be used to override a system header
4990 file, substituting your own version, since these directories are
4991 searched before the system header file directories. However, you should
4992 not use this option to add directories that contain vendor-supplied
4993 system header files (use @option{-isystem} for that). If you use more than
4994 one @option{-I} option, the directories are scanned in left-to-right
4995 order; the standard system directories come after.
4997 If a standard system include directory, or a directory specified with
4998 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
4999 option will be ignored. The directory will still be searched but as a
5000 system directory at its normal position in the system include chain.
5001 This is to ensure that GCC's procedure to fix buggy system headers and
5002 the ordering for the include_next directive are not inadvertently changed.
5003 If you really need to change the search order for system directories,
5004 use the @option{-nostdinc} and/or @option{-isystem} options.
5008 Any directories you specify with @option{-I} options before the @option{-I-}
5009 option are searched only for the case of @samp{#include "@var{file}"};
5010 they are not searched for @samp{#include <@var{file}>}.
5012 If additional directories are specified with @option{-I} options after
5013 the @option{-I-}, these directories are searched for all @samp{#include}
5014 directives. (Ordinarily @emph{all} @option{-I} directories are used
5017 In addition, the @option{-I-} option inhibits the use of the current
5018 directory (where the current input file came from) as the first search
5019 directory for @samp{#include "@var{file}"}. There is no way to
5020 override this effect of @option{-I-}. With @option{-I.} you can specify
5021 searching the directory which was current when the compiler was
5022 invoked. That is not exactly the same as what the preprocessor does
5023 by default, but it is often satisfactory.
5025 @option{-I-} does not inhibit the use of the standard system directories
5026 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5031 Add directory @var{dir} to the list of directories to be searched
5034 @item -B@var{prefix}
5036 This option specifies where to find the executables, libraries,
5037 include files, and data files of the compiler itself.
5039 The compiler driver program runs one or more of the subprograms
5040 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5041 @var{prefix} as a prefix for each program it tries to run, both with and
5042 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5044 For each subprogram to be run, the compiler driver first tries the
5045 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5046 was not specified, the driver tries two standard prefixes, which are
5047 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
5048 those results in a file name that is found, the unmodified program
5049 name is searched for using the directories specified in your
5050 @env{PATH} environment variable.
5052 The compiler will check to see if the path provided by the @option{-B}
5053 refers to a directory, and if necessary it will add a directory
5054 separator character at the end of the path.
5056 @option{-B} prefixes that effectively specify directory names also apply
5057 to libraries in the linker, because the compiler translates these
5058 options into @option{-L} options for the linker. They also apply to
5059 includes files in the preprocessor, because the compiler translates these
5060 options into @option{-isystem} options for the preprocessor. In this case,
5061 the compiler appends @samp{include} to the prefix.
5063 The run-time support file @file{libgcc.a} can also be searched for using
5064 the @option{-B} prefix, if needed. If it is not found there, the two
5065 standard prefixes above are tried, and that is all. The file is left
5066 out of the link if it is not found by those means.
5068 Another way to specify a prefix much like the @option{-B} prefix is to use
5069 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5072 As a special kludge, if the path provided by @option{-B} is
5073 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5074 9, then it will be replaced by @file{[dir/]include}. This is to help
5075 with boot-strapping the compiler.
5077 @item -specs=@var{file}
5079 Process @var{file} after the compiler reads in the standard @file{specs}
5080 file, in order to override the defaults that the @file{gcc} driver
5081 program uses when determining what switches to pass to @file{cc1},
5082 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5083 @option{-specs=@var{file}} can be specified on the command line, and they
5084 are processed in order, from left to right.
5090 @section Specifying subprocesses and the switches to pass to them
5093 @command{gcc} is a driver program. It performs its job by invoking a
5094 sequence of other programs to do the work of compiling, assembling and
5095 linking. GCC interprets its command-line parameters and uses these to
5096 deduce which programs it should invoke, and which command-line options
5097 it ought to place on their command lines. This behavior is controlled
5098 by @dfn{spec strings}. In most cases there is one spec string for each
5099 program that GCC can invoke, but a few programs have multiple spec
5100 strings to control their behavior. The spec strings built into GCC can
5101 be overridden by using the @option{-specs=} command-line switch to specify
5104 @dfn{Spec files} are plaintext files that are used to construct spec
5105 strings. They consist of a sequence of directives separated by blank
5106 lines. The type of directive is determined by the first non-whitespace
5107 character on the line and it can be one of the following:
5110 @item %@var{command}
5111 Issues a @var{command} to the spec file processor. The commands that can
5115 @item %include <@var{file}>
5117 Search for @var{file} and insert its text at the current point in the
5120 @item %include_noerr <@var{file}>
5121 @cindex %include_noerr
5122 Just like @samp{%include}, but do not generate an error message if the include
5123 file cannot be found.
5125 @item %rename @var{old_name} @var{new_name}
5127 Rename the spec string @var{old_name} to @var{new_name}.
5131 @item *[@var{spec_name}]:
5132 This tells the compiler to create, override or delete the named spec
5133 string. All lines after this directive up to the next directive or
5134 blank line are considered to be the text for the spec string. If this
5135 results in an empty string then the spec will be deleted. (Or, if the
5136 spec did not exist, then nothing will happened.) Otherwise, if the spec
5137 does not currently exist a new spec will be created. If the spec does
5138 exist then its contents will be overridden by the text of this
5139 directive, unless the first character of that text is the @samp{+}
5140 character, in which case the text will be appended to the spec.
5142 @item [@var{suffix}]:
5143 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5144 and up to the next directive or blank line are considered to make up the
5145 spec string for the indicated suffix. When the compiler encounters an
5146 input file with the named suffix, it will processes the spec string in
5147 order to work out how to compile that file. For example:
5154 This says that any input file whose name ends in @samp{.ZZ} should be
5155 passed to the program @samp{z-compile}, which should be invoked with the
5156 command-line switch @option{-input} and with the result of performing the
5157 @samp{%i} substitution. (See below.)
5159 As an alternative to providing a spec string, the text that follows a
5160 suffix directive can be one of the following:
5163 @item @@@var{language}
5164 This says that the suffix is an alias for a known @var{language}. This is
5165 similar to using the @option{-x} command-line switch to GCC to specify a
5166 language explicitly. For example:
5173 Says that .ZZ files are, in fact, C++ source files.
5176 This causes an error messages saying:
5179 @var{name} compiler not installed on this system.
5183 GCC already has an extensive list of suffixes built into it.
5184 This directive will add an entry to the end of the list of suffixes, but
5185 since the list is searched from the end backwards, it is effectively
5186 possible to override earlier entries using this technique.
5190 GCC has the following spec strings built into it. Spec files can
5191 override these strings or create their own. Note that individual
5192 targets can also add their own spec strings to this list.
5195 asm Options to pass to the assembler
5196 asm_final Options to pass to the assembler post-processor
5197 cpp Options to pass to the C preprocessor
5198 cc1 Options to pass to the C compiler
5199 cc1plus Options to pass to the C++ compiler
5200 endfile Object files to include at the end of the link
5201 link Options to pass to the linker
5202 lib Libraries to include on the command line to the linker
5203 libgcc Decides which GCC support library to pass to the linker
5204 linker Sets the name of the linker
5205 predefines Defines to be passed to the C preprocessor
5206 signed_char Defines to pass to CPP to say whether @code{char} is signed
5208 startfile Object files to include at the start of the link
5211 Here is a small example of a spec file:
5217 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5220 This example renames the spec called @samp{lib} to @samp{old_lib} and
5221 then overrides the previous definition of @samp{lib} with a new one.
5222 The new definition adds in some extra command-line options before
5223 including the text of the old definition.
5225 @dfn{Spec strings} are a list of command-line options to be passed to their
5226 corresponding program. In addition, the spec strings can contain
5227 @samp{%}-prefixed sequences to substitute variable text or to
5228 conditionally insert text into the command line. Using these constructs
5229 it is possible to generate quite complex command lines.
5231 Here is a table of all defined @samp{%}-sequences for spec
5232 strings. Note that spaces are not generated automatically around the
5233 results of expanding these sequences. Therefore you can concatenate them
5234 together or combine them with constant text in a single argument.
5238 Substitute one @samp{%} into the program name or argument.
5241 Substitute the name of the input file being processed.
5244 Substitute the basename of the input file being processed.
5245 This is the substring up to (and not including) the last period
5246 and not including the directory.
5249 This is the same as @samp{%b}, but include the file suffix (text after
5253 Marks the argument containing or following the @samp{%d} as a
5254 temporary file name, so that that file will be deleted if GCC exits
5255 successfully. Unlike @samp{%g}, this contributes no text to the
5258 @item %g@var{suffix}
5259 Substitute a file name that has suffix @var{suffix} and is chosen
5260 once per compilation, and mark the argument in the same way as
5261 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5262 name is now chosen in a way that is hard to predict even when previously
5263 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5264 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5265 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5266 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5267 was simply substituted with a file name chosen once per compilation,
5268 without regard to any appended suffix (which was therefore treated
5269 just like ordinary text), making such attacks more likely to succeed.
5271 @item %u@var{suffix}
5272 Like @samp{%g}, but generates a new temporary file name even if
5273 @samp{%u@var{suffix}} was already seen.
5275 @item %U@var{suffix}
5276 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5277 new one if there is no such last file name. In the absence of any
5278 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5279 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5280 would involve the generation of two distinct file names, one
5281 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5282 simply substituted with a file name chosen for the previous @samp{%u},
5283 without regard to any appended suffix.
5285 @item %j@var{suffix}
5286 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5287 writable, and if save-temps is off; otherwise, substitute the name
5288 of a temporary file, just like @samp{%u}. This temporary file is not
5289 meant for communication between processes, but rather as a junk
5292 @item %|@var{suffix}
5293 @itemx %m@var{suffix}
5294 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5295 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5296 all. These are the two most common ways to instruct a program that it
5297 should read from standard input or write to standard output. If you
5298 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5299 construct: see for example @file{f/lang-specs.h}.
5301 @item %.@var{SUFFIX}
5302 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5303 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5304 terminated by the next space or %.
5307 Marks the argument containing or following the @samp{%w} as the
5308 designated output file of this compilation. This puts the argument
5309 into the sequence of arguments that @samp{%o} will substitute later.
5312 Substitutes the names of all the output files, with spaces
5313 automatically placed around them. You should write spaces
5314 around the @samp{%o} as well or the results are undefined.
5315 @samp{%o} is for use in the specs for running the linker.
5316 Input files whose names have no recognized suffix are not compiled
5317 at all, but they are included among the output files, so they will
5321 Substitutes the suffix for object files. Note that this is
5322 handled specially when it immediately follows @samp{%g, %u, or %U},
5323 because of the need for those to form complete file names. The
5324 handling is such that @samp{%O} is treated exactly as if it had already
5325 been substituted, except that @samp{%g, %u, and %U} do not currently
5326 support additional @var{suffix} characters following @samp{%O} as they would
5327 following, for example, @samp{.o}.
5330 Substitutes the standard macro predefinitions for the
5331 current target machine. Use this when running @code{cpp}.
5334 Like @samp{%p}, but puts @samp{__} before and after the name of each
5335 predefined macro, except for macros that start with @samp{__} or with
5336 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5340 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5341 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5342 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5346 Current argument is the name of a library or startup file of some sort.
5347 Search for that file in a standard list of directories and substitute
5348 the full name found.
5351 Print @var{str} as an error message. @var{str} is terminated by a newline.
5352 Use this when inconsistent options are detected.
5355 Substitute the contents of spec string @var{name} at this point.
5358 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5360 @item %x@{@var{option}@}
5361 Accumulate an option for @samp{%X}.
5364 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5368 Output the accumulated assembler options specified by @option{-Wa}.
5371 Output the accumulated preprocessor options specified by @option{-Wp}.
5374 Process the @code{asm} spec. This is used to compute the
5375 switches to be passed to the assembler.
5378 Process the @code{asm_final} spec. This is a spec string for
5379 passing switches to an assembler post-processor, if such a program is
5383 Process the @code{link} spec. This is the spec for computing the
5384 command line passed to the linker. Typically it will make use of the
5385 @samp{%L %G %S %D and %E} sequences.
5388 Dump out a @option{-L} option for each directory that GCC believes might
5389 contain startup files. If the target supports multilibs then the
5390 current multilib directory will be prepended to each of these paths.
5393 Output the multilib directory with directory separators replaced with
5394 @samp{_}. If multilib directories are not set, or the multilib directory is
5395 @file{.} then this option emits nothing.
5398 Process the @code{lib} spec. This is a spec string for deciding which
5399 libraries should be included on the command line to the linker.
5402 Process the @code{libgcc} spec. This is a spec string for deciding
5403 which GCC support library should be included on the command line to the linker.
5406 Process the @code{startfile} spec. This is a spec for deciding which
5407 object files should be the first ones passed to the linker. Typically
5408 this might be a file named @file{crt0.o}.
5411 Process the @code{endfile} spec. This is a spec string that specifies
5412 the last object files that will be passed to the linker.
5415 Process the @code{cpp} spec. This is used to construct the arguments
5416 to be passed to the C preprocessor.
5419 Process the @code{signed_char} spec. This is intended to be used
5420 to tell cpp whether a char is signed. It typically has the definition:
5422 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5426 Process the @code{cc1} spec. This is used to construct the options to be
5427 passed to the actual C compiler (@samp{cc1}).
5430 Process the @code{cc1plus} spec. This is used to construct the options to be
5431 passed to the actual C++ compiler (@samp{cc1plus}).
5434 Substitute the variable part of a matched option. See below.
5435 Note that each comma in the substituted string is replaced by
5439 Remove all occurrences of @code{-S} from the command line. Note---this
5440 command is position dependent. @samp{%} commands in the spec string
5441 before this one will see @code{-S}, @samp{%} commands in the spec string
5442 after this one will not.
5444 @item %:@var{function}(@var{args})
5445 Call the named function @var{function}, passing it @var{args}.
5446 @var{args} is first processed as a nested spec string, then split
5447 into an argument vector in the usual fashion. The function returns
5448 a string which is processed as if it had appeared literally as part
5449 of the current spec.
5451 The following built-in spec functions are provided:
5454 @item @code{if-exists}
5455 The @code{if-exists} spec function takes one argument, an absolute
5456 pathname to a file. If the file exists, @code{if-exists} returns the
5457 pathname. Here is a small example of its usage:
5461 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5464 @item @code{if-exists-else}
5465 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5466 spec function, except that it takes two arguments. The first argument is
5467 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5468 returns the pathname. If it does not exist, it returns the second argument.
5469 This way, @code{if-exists-else} can be used to select one file or another,
5470 based on the existence of the first. Here is a small example of its usage:
5474 crt0%O%s %:if-exists(crti%O%s) \
5475 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5480 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5481 If that switch was not specified, this substitutes nothing. Note that
5482 the leading dash is omitted when specifying this option, and it is
5483 automatically inserted if the substitution is performed. Thus the spec
5484 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5485 and would output the command line option @option{-foo}.
5487 @item %W@{@code{S}@}
5488 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5491 @item %@{@code{S}*@}
5492 Substitutes all the switches specified to GCC whose names start
5493 with @code{-S}, but which also take an argument. This is used for
5494 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5495 GCC considers @option{-o foo} as being
5496 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5497 text, including the space. Thus two arguments would be generated.
5499 @item %@{@code{S}*&@code{T}*@}
5500 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5501 (the order of @code{S} and @code{T} in the spec is not significant).
5502 There can be any number of ampersand-separated variables; for each the
5503 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5505 @item %@{@code{S}:@code{X}@}
5506 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5508 @item %@{!@code{S}:@code{X}@}
5509 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5511 @item %@{@code{S}*:@code{X}@}
5512 Substitutes @code{X} if one or more switches whose names start with
5513 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5514 once, no matter how many such switches appeared. However, if @code{%*}
5515 appears somewhere in @code{X}, then @code{X} will be substituted once
5516 for each matching switch, with the @code{%*} replaced by the part of
5517 that switch that matched the @code{*}.
5519 @item %@{.@code{S}:@code{X}@}
5520 Substitutes @code{X}, if processing a file with suffix @code{S}.
5522 @item %@{!.@code{S}:@code{X}@}
5523 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5525 @item %@{@code{S}|@code{P}:@code{X}@}
5526 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5527 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5528 although they have a stronger binding than the @samp{|}. If @code{%*}
5529 appears in @code{X}, all of the alternatives must be starred, and only
5530 the first matching alternative is substituted.
5532 For example, a spec string like this:
5535 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5538 will output the following command-line options from the following input
5539 command-line options:
5544 -d fred.c -foo -baz -boggle
5545 -d jim.d -bar -baz -boggle
5548 @item %@{S:X; T:Y; :D@}
5550 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5551 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5552 be as many clauses as you need. This may be combined with @code{.},
5553 @code{!}, @code{|}, and @code{*} as needed.
5558 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5559 construct may contain other nested @samp{%} constructs or spaces, or
5560 even newlines. They are processed as usual, as described above.
5561 Trailing white space in @code{X} is ignored. White space may also
5562 appear anywhere on the left side of the colon in these constructs,
5563 except between @code{.} or @code{*} and the corresponding word.
5565 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5566 handled specifically in these constructs. If another value of
5567 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5568 @option{-W} switch is found later in the command line, the earlier
5569 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5570 just one letter, which passes all matching options.
5572 The character @samp{|} at the beginning of the predicate text is used to
5573 indicate that a command should be piped to the following command, but
5574 only if @option{-pipe} is specified.
5576 It is built into GCC which switches take arguments and which do not.
5577 (You might think it would be useful to generalize this to allow each
5578 compiler's spec to say which switches take arguments. But this cannot
5579 be done in a consistent fashion. GCC cannot even decide which input
5580 files have been specified without knowing which switches take arguments,
5581 and it must know which input files to compile in order to tell which
5584 GCC also knows implicitly that arguments starting in @option{-l} are to be
5585 treated as compiler output files, and passed to the linker in their
5586 proper position among the other output files.
5588 @c man begin OPTIONS
5590 @node Target Options
5591 @section Specifying Target Machine and Compiler Version
5592 @cindex target options
5593 @cindex cross compiling
5594 @cindex specifying machine version
5595 @cindex specifying compiler version and target machine
5596 @cindex compiler version, specifying
5597 @cindex target machine, specifying
5599 The usual way to run GCC is to run the executable called @file{gcc}, or
5600 @file{<machine>-gcc} when cross-compiling, or
5601 @file{<machine>-gcc-<version>} to run a version other than the one that
5602 was installed last. Sometimes this is inconvenient, so GCC provides
5603 options that will switch to another cross-compiler or version.
5606 @item -b @var{machine}
5608 The argument @var{machine} specifies the target machine for compilation.
5610 The value to use for @var{machine} is the same as was specified as the
5611 machine type when configuring GCC as a cross-compiler. For
5612 example, if a cross-compiler was configured with @samp{configure
5613 i386v}, meaning to compile for an 80386 running System V, then you
5614 would specify @option{-b i386v} to run that cross compiler.
5616 @item -V @var{version}
5618 The argument @var{version} specifies which version of GCC to run.
5619 This is useful when multiple versions are installed. For example,
5620 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5623 The @option{-V} and @option{-b} options work by running the
5624 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5625 use them if you can just run that directly.
5627 @node Submodel Options
5628 @section Hardware Models and Configurations
5629 @cindex submodel options
5630 @cindex specifying hardware config
5631 @cindex hardware models and configurations, specifying
5632 @cindex machine dependent options
5634 Earlier we discussed the standard option @option{-b} which chooses among
5635 different installed compilers for completely different target
5636 machines, such as VAX vs.@: 68000 vs.@: 80386.
5638 In addition, each of these target machine types can have its own
5639 special options, starting with @samp{-m}, to choose among various
5640 hardware models or configurations---for example, 68010 vs 68020,
5641 floating coprocessor or none. A single installed version of the
5642 compiler can compile for any model or configuration, according to the
5645 Some configurations of the compiler also support additional special
5646 options, usually for compatibility with other compilers on the same
5649 These options are defined by the macro @code{TARGET_SWITCHES} in the
5650 machine description. The default for the options is also defined by
5651 that macro, which enables you to change the defaults.
5663 * RS/6000 and PowerPC Options::
5667 * i386 and x86-64 Options::
5669 * Intel 960 Options::
5670 * DEC Alpha Options::
5671 * DEC Alpha/VMS Options::
5674 * System V Options::
5675 * TMS320C3x/C4x Options::
5683 * S/390 and zSeries Options::
5687 * Xstormy16 Options::
5692 @node M680x0 Options
5693 @subsection M680x0 Options
5694 @cindex M680x0 options
5696 These are the @samp{-m} options defined for the 68000 series. The default
5697 values for these options depends on which style of 68000 was selected when
5698 the compiler was configured; the defaults for the most common choices are
5706 Generate output for a 68000. This is the default
5707 when the compiler is configured for 68000-based systems.
5709 Use this option for microcontrollers with a 68000 or EC000 core,
5710 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5716 Generate output for a 68020. This is the default
5717 when the compiler is configured for 68020-based systems.
5721 Generate output containing 68881 instructions for floating point.
5722 This is the default for most 68020 systems unless @option{--nfp} was
5723 specified when the compiler was configured.
5727 Generate output for a 68030. This is the default when the compiler is
5728 configured for 68030-based systems.
5732 Generate output for a 68040. This is the default when the compiler is
5733 configured for 68040-based systems.
5735 This option inhibits the use of 68881/68882 instructions that have to be
5736 emulated by software on the 68040. Use this option if your 68040 does not
5737 have code to emulate those instructions.
5741 Generate output for a 68060. This is the default when the compiler is
5742 configured for 68060-based systems.
5744 This option inhibits the use of 68020 and 68881/68882 instructions that
5745 have to be emulated by software on the 68060. Use this option if your 68060
5746 does not have code to emulate those instructions.
5750 Generate output for a CPU32. This is the default
5751 when the compiler is configured for CPU32-based systems.
5753 Use this option for microcontrollers with a
5754 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5755 68336, 68340, 68341, 68349 and 68360.
5759 Generate output for a 520X ``coldfire'' family cpu. This is the default
5760 when the compiler is configured for 520X-based systems.
5762 Use this option for microcontroller with a 5200 core, including
5763 the MCF5202, MCF5203, MCF5204 and MCF5202.
5768 Generate output for a 68040, without using any of the new instructions.
5769 This results in code which can run relatively efficiently on either a
5770 68020/68881 or a 68030 or a 68040. The generated code does use the
5771 68881 instructions that are emulated on the 68040.
5775 Generate output for a 68060, without using any of the new instructions.
5776 This results in code which can run relatively efficiently on either a
5777 68020/68881 or a 68030 or a 68040. The generated code does use the
5778 68881 instructions that are emulated on the 68060.
5781 @opindex msoft-float
5782 Generate output containing library calls for floating point.
5783 @strong{Warning:} the requisite libraries are not available for all m68k
5784 targets. Normally the facilities of the machine's usual C compiler are
5785 used, but this can't be done directly in cross-compilation. You must
5786 make your own arrangements to provide suitable library functions for
5787 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5788 @samp{m68k-*-coff} do provide software floating point support.
5792 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5795 @opindex mnobitfield
5796 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5797 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5801 Do use the bit-field instructions. The @option{-m68020} option implies
5802 @option{-mbitfield}. This is the default if you use a configuration
5803 designed for a 68020.
5807 Use a different function-calling convention, in which functions
5808 that take a fixed number of arguments return with the @code{rtd}
5809 instruction, which pops their arguments while returning. This
5810 saves one instruction in the caller since there is no need to pop
5811 the arguments there.
5813 This calling convention is incompatible with the one normally
5814 used on Unix, so you cannot use it if you need to call libraries
5815 compiled with the Unix compiler.
5817 Also, you must provide function prototypes for all functions that
5818 take variable numbers of arguments (including @code{printf});
5819 otherwise incorrect code will be generated for calls to those
5822 In addition, seriously incorrect code will result if you call a
5823 function with too many arguments. (Normally, extra arguments are
5824 harmlessly ignored.)
5826 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5827 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5830 @itemx -mno-align-int
5832 @opindex mno-align-int
5833 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5834 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5835 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5836 Aligning variables on 32-bit boundaries produces code that runs somewhat
5837 faster on processors with 32-bit busses at the expense of more memory.
5839 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5840 align structures containing the above types differently than
5841 most published application binary interface specifications for the m68k.
5845 Use the pc-relative addressing mode of the 68000 directly, instead of
5846 using a global offset table. At present, this option implies @option{-fpic},
5847 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5848 not presently supported with @option{-mpcrel}, though this could be supported for
5849 68020 and higher processors.
5851 @item -mno-strict-align
5852 @itemx -mstrict-align
5853 @opindex mno-strict-align
5854 @opindex mstrict-align
5855 Do not (do) assume that unaligned memory references will be handled by
5860 @node M68hc1x Options
5861 @subsection M68hc1x Options
5862 @cindex M68hc1x options
5864 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5865 microcontrollers. The default values for these options depends on
5866 which style of microcontroller was selected when the compiler was configured;
5867 the defaults for the most common choices are given below.
5874 Generate output for a 68HC11. This is the default
5875 when the compiler is configured for 68HC11-based systems.
5881 Generate output for a 68HC12. This is the default
5882 when the compiler is configured for 68HC12-based systems.
5888 Generate output for a 68HCS12.
5891 @opindex mauto-incdec
5892 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5899 Enable the use of 68HC12 min and max instructions.
5902 @itemx -mno-long-calls
5903 @opindex mlong-calls
5904 @opindex mno-long-calls
5905 Treat all calls as being far away (near). If calls are assumed to be
5906 far away, the compiler will use the @code{call} instruction to
5907 call a function and the @code{rtc} instruction for returning.
5911 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5913 @item -msoft-reg-count=@var{count}
5914 @opindex msoft-reg-count
5915 Specify the number of pseudo-soft registers which are used for the
5916 code generation. The maximum number is 32. Using more pseudo-soft
5917 register may or may not result in better code depending on the program.
5918 The default is 4 for 68HC11 and 2 for 68HC12.
5923 @subsection VAX Options
5926 These @samp{-m} options are defined for the VAX:
5931 Do not output certain jump instructions (@code{aobleq} and so on)
5932 that the Unix assembler for the VAX cannot handle across long
5937 Do output those jump instructions, on the assumption that you
5938 will assemble with the GNU assembler.
5942 Output code for g-format floating point numbers instead of d-format.
5946 @subsection SPARC Options
5947 @cindex SPARC options
5949 These @samp{-m} switches are supported on the SPARC:
5954 @opindex mno-app-regs
5956 Specify @option{-mapp-regs} to generate output using the global registers
5957 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5960 To be fully SVR4 ABI compliant at the cost of some performance loss,
5961 specify @option{-mno-app-regs}. You should compile libraries and system
5962 software with this option.
5967 @opindex mhard-float
5968 Generate output containing floating point instructions. This is the
5974 @opindex msoft-float
5975 Generate output containing library calls for floating point.
5976 @strong{Warning:} the requisite libraries are not available for all SPARC
5977 targets. Normally the facilities of the machine's usual C compiler are
5978 used, but this cannot be done directly in cross-compilation. You must make
5979 your own arrangements to provide suitable library functions for
5980 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5981 @samp{sparclite-*-*} do provide software floating point support.
5983 @option{-msoft-float} changes the calling convention in the output file;
5984 therefore, it is only useful if you compile @emph{all} of a program with
5985 this option. In particular, you need to compile @file{libgcc.a}, the
5986 library that comes with GCC, with @option{-msoft-float} in order for
5989 @item -mhard-quad-float
5990 @opindex mhard-quad-float
5991 Generate output containing quad-word (long double) floating point
5995 @opindex mimpure-text
5996 @option{-mimpure-text}, used in addition to @option{-shared}, tells
5997 the compiler to not pass @option{-z text} to the linker when linking a
5998 shared object. Using this option, you can link position-dependent
5999 code into a shared object.
6001 @option{-mimpure-text} suppresses the ``relocations remain against
6002 allocatable but non-writable sections'' linker error message.
6003 However, the necessary relocations will trigger copy-on-write, and the
6004 shared object is not actually shared across processes. Instead of
6005 using @option{-mimpure-text}, you should compile all source code with
6006 @option{-fpic} or @option{-fPIC}.
6008 This option is only available on SunOS and Solaris.
6010 @item -msoft-quad-float
6011 @opindex msoft-quad-float
6012 Generate output containing library calls for quad-word (long double)
6013 floating point instructions. The functions called are those specified
6014 in the SPARC ABI@. This is the default.
6016 As of this writing, there are no sparc implementations that have hardware
6017 support for the quad-word floating point instructions. They all invoke
6018 a trap handler for one of these instructions, and then the trap handler
6019 emulates the effect of the instruction. Because of the trap handler overhead,
6020 this is much slower than calling the ABI library routines. Thus the
6021 @option{-msoft-quad-float} option is the default.
6027 With @option{-mflat}, the compiler does not generate save/restore instructions
6028 and will use a ``flat'' or single register window calling convention.
6029 This model uses %i7 as the frame pointer and is compatible with the normal
6030 register window model. Code from either may be intermixed.
6031 The local registers and the input registers (0--5) are still treated as
6032 ``call saved'' registers and will be saved on the stack as necessary.
6034 With @option{-mno-flat} (the default), the compiler emits save/restore
6035 instructions (except for leaf functions) and is the normal mode of operation.
6037 @item -mno-unaligned-doubles
6038 @itemx -munaligned-doubles
6039 @opindex mno-unaligned-doubles
6040 @opindex munaligned-doubles
6041 Assume that doubles have 8 byte alignment. This is the default.
6043 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6044 alignment only if they are contained in another type, or if they have an
6045 absolute address. Otherwise, it assumes they have 4 byte alignment.
6046 Specifying this option avoids some rare compatibility problems with code
6047 generated by other compilers. It is not the default because it results
6048 in a performance loss, especially for floating point code.
6050 @item -mno-faster-structs
6051 @itemx -mfaster-structs
6052 @opindex mno-faster-structs
6053 @opindex mfaster-structs
6054 With @option{-mfaster-structs}, the compiler assumes that structures
6055 should have 8 byte alignment. This enables the use of pairs of
6056 @code{ldd} and @code{std} instructions for copies in structure
6057 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6058 However, the use of this changed alignment directly violates the SPARC
6059 ABI@. Thus, it's intended only for use on targets where the developer
6060 acknowledges that their resulting code will not be directly in line with
6061 the rules of the ABI@.
6067 These two options select variations on the SPARC architecture.
6069 By default (unless specifically configured for the Fujitsu SPARClite),
6070 GCC generates code for the v7 variant of the SPARC architecture.
6072 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6073 code is that the compiler emits the integer multiply and integer
6074 divide instructions which exist in SPARC v8 but not in SPARC v7.
6076 @option{-msparclite} will give you SPARClite code. This adds the integer
6077 multiply, integer divide step and scan (@code{ffs}) instructions which
6078 exist in SPARClite but not in SPARC v7.
6080 These options are deprecated and will be deleted in a future GCC release.
6081 They have been replaced with @option{-mcpu=xxx}.
6086 @opindex msupersparc
6087 These two options select the processor for which the code is optimized.
6089 With @option{-mcypress} (the default), the compiler optimizes code for the
6090 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6091 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6093 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6094 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6095 of the full SPARC v8 instruction set.
6097 These options are deprecated and will be deleted in a future GCC release.
6098 They have been replaced with @option{-mcpu=xxx}.
6100 @item -mcpu=@var{cpu_type}
6102 Set the instruction set, register set, and instruction scheduling parameters
6103 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6104 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6105 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6106 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6109 Default instruction scheduling parameters are used for values that select
6110 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6111 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6113 Here is a list of each supported architecture and their supported
6118 v8: supersparc, hypersparc
6119 sparclite: f930, f934, sparclite86x
6121 v9: ultrasparc, ultrasparc3
6124 @item -mtune=@var{cpu_type}
6126 Set the instruction scheduling parameters for machine type
6127 @var{cpu_type}, but do not set the instruction set or register set that the
6128 option @option{-mcpu=@var{cpu_type}} would.
6130 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6131 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6132 that select a particular cpu implementation. Those are @samp{cypress},
6133 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6134 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6139 These @samp{-m} switches are supported in addition to the above
6140 on the SPARCLET processor.
6143 @item -mlittle-endian
6144 @opindex mlittle-endian
6145 Generate code for a processor running in little-endian mode.
6149 Treat register @code{%g0} as a normal register.
6150 GCC will continue to clobber it as necessary but will not assume
6151 it always reads as 0.
6153 @item -mbroken-saverestore
6154 @opindex mbroken-saverestore
6155 Generate code that does not use non-trivial forms of the @code{save} and
6156 @code{restore} instructions. Early versions of the SPARCLET processor do
6157 not correctly handle @code{save} and @code{restore} instructions used with
6158 arguments. They correctly handle them used without arguments. A @code{save}
6159 instruction used without arguments increments the current window pointer
6160 but does not allocate a new stack frame. It is assumed that the window
6161 overflow trap handler will properly handle this case as will interrupt
6165 These @samp{-m} switches are supported in addition to the above
6166 on SPARC V9 processors in 64-bit environments.
6169 @item -mlittle-endian
6170 @opindex mlittle-endian
6171 Generate code for a processor running in little-endian mode.
6177 Generate code for a 32-bit or 64-bit environment.
6178 The 32-bit environment sets int, long and pointer to 32 bits.
6179 The 64-bit environment sets int to 32 bits and long and pointer
6182 @item -mcmodel=medlow
6183 @opindex mcmodel=medlow
6184 Generate code for the Medium/Low code model: the program must be linked
6185 in the low 32 bits of the address space. Pointers are 64 bits.
6186 Programs can be statically or dynamically linked.
6188 @item -mcmodel=medmid
6189 @opindex mcmodel=medmid
6190 Generate code for the Medium/Middle code model: the program must be linked
6191 in the low 44 bits of the address space, the text segment must be less than
6192 2G bytes, and data segment must be within 2G of the text segment.
6193 Pointers are 64 bits.
6195 @item -mcmodel=medany
6196 @opindex mcmodel=medany
6197 Generate code for the Medium/Anywhere code model: the program may be linked
6198 anywhere in the address space, the text segment must be less than
6199 2G bytes, and data segment must be within 2G of the text segment.
6200 Pointers are 64 bits.
6202 @item -mcmodel=embmedany
6203 @opindex mcmodel=embmedany
6204 Generate code for the Medium/Anywhere code model for embedded systems:
6205 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6206 (determined at link time). Register %g4 points to the base of the
6207 data segment. Pointers are still 64 bits.
6208 Programs are statically linked, PIC is not supported.
6211 @itemx -mno-stack-bias
6212 @opindex mstack-bias
6213 @opindex mno-stack-bias
6214 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6215 frame pointer if present, are offset by @minus{}2047 which must be added back
6216 when making stack frame references.
6217 Otherwise, assume no such offset is present.
6221 @subsection ARM Options
6224 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6229 @opindex mapcs-frame
6230 Generate a stack frame that is compliant with the ARM Procedure Call
6231 Standard for all functions, even if this is not strictly necessary for
6232 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6233 with this option will cause the stack frames not to be generated for
6234 leaf functions. The default is @option{-mno-apcs-frame}.
6238 This is a synonym for @option{-mapcs-frame}.
6242 Generate code for a processor running with a 26-bit program counter,
6243 and conforming to the function calling standards for the APCS 26-bit
6244 option. This option replaces the @option{-m2} and @option{-m3} options
6245 of previous releases of the compiler.
6249 Generate code for a processor running with a 32-bit program counter,
6250 and conforming to the function calling standards for the APCS 32-bit
6251 option. This option replaces the @option{-m6} option of previous releases
6255 @c not currently implemented
6256 @item -mapcs-stack-check
6257 @opindex mapcs-stack-check
6258 Generate code to check the amount of stack space available upon entry to
6259 every function (that actually uses some stack space). If there is
6260 insufficient space available then either the function
6261 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6262 called, depending upon the amount of stack space required. The run time
6263 system is required to provide these functions. The default is
6264 @option{-mno-apcs-stack-check}, since this produces smaller code.
6266 @c not currently implemented
6268 @opindex mapcs-float
6269 Pass floating point arguments using the float point registers. This is
6270 one of the variants of the APCS@. This option is recommended if the
6271 target hardware has a floating point unit or if a lot of floating point
6272 arithmetic is going to be performed by the code. The default is
6273 @option{-mno-apcs-float}, since integer only code is slightly increased in
6274 size if @option{-mapcs-float} is used.
6276 @c not currently implemented
6277 @item -mapcs-reentrant
6278 @opindex mapcs-reentrant
6279 Generate reentrant, position independent code. The default is
6280 @option{-mno-apcs-reentrant}.
6283 @item -mthumb-interwork
6284 @opindex mthumb-interwork
6285 Generate code which supports calling between the ARM and Thumb
6286 instruction sets. Without this option the two instruction sets cannot
6287 be reliably used inside one program. The default is
6288 @option{-mno-thumb-interwork}, since slightly larger code is generated
6289 when @option{-mthumb-interwork} is specified.
6291 @item -mno-sched-prolog
6292 @opindex mno-sched-prolog
6293 Prevent the reordering of instructions in the function prolog, or the
6294 merging of those instruction with the instructions in the function's
6295 body. This means that all functions will start with a recognizable set
6296 of instructions (or in fact one of a choice from a small set of
6297 different function prologues), and this information can be used to
6298 locate the start if functions inside an executable piece of code. The
6299 default is @option{-msched-prolog}.
6302 @opindex mhard-float
6303 Generate output containing floating point instructions. This is the
6307 @opindex msoft-float
6308 Generate output containing library calls for floating point.
6309 @strong{Warning:} the requisite libraries are not available for all ARM
6310 targets. Normally the facilities of the machine's usual C compiler are
6311 used, but this cannot be done directly in cross-compilation. You must make
6312 your own arrangements to provide suitable library functions for
6315 @option{-msoft-float} changes the calling convention in the output file;
6316 therefore, it is only useful if you compile @emph{all} of a program with
6317 this option. In particular, you need to compile @file{libgcc.a}, the
6318 library that comes with GCC, with @option{-msoft-float} in order for
6321 @item -mlittle-endian
6322 @opindex mlittle-endian
6323 Generate code for a processor running in little-endian mode. This is
6324 the default for all standard configurations.
6327 @opindex mbig-endian
6328 Generate code for a processor running in big-endian mode; the default is
6329 to compile code for a little-endian processor.
6331 @item -mwords-little-endian
6332 @opindex mwords-little-endian
6333 This option only applies when generating code for big-endian processors.
6334 Generate code for a little-endian word order but a big-endian byte
6335 order. That is, a byte order of the form @samp{32107654}. Note: this
6336 option should only be used if you require compatibility with code for
6337 big-endian ARM processors generated by versions of the compiler prior to
6340 @item -malignment-traps
6341 @opindex malignment-traps
6342 Generate code that will not trap if the MMU has alignment traps enabled.
6343 On ARM architectures prior to ARMv4, there were no instructions to
6344 access half-word objects stored in memory. However, when reading from
6345 memory a feature of the ARM architecture allows a word load to be used,
6346 even if the address is unaligned, and the processor core will rotate the
6347 data as it is being loaded. This option tells the compiler that such
6348 misaligned accesses will cause a MMU trap and that it should instead
6349 synthesize the access as a series of byte accesses. The compiler can
6350 still use word accesses to load half-word data if it knows that the
6351 address is aligned to a word boundary.
6353 This option is ignored when compiling for ARM architecture 4 or later,
6354 since these processors have instructions to directly access half-word
6357 @item -mno-alignment-traps
6358 @opindex mno-alignment-traps
6359 Generate code that assumes that the MMU will not trap unaligned
6360 accesses. This produces better code when the target instruction set
6361 does not have half-word memory operations (i.e.@: implementations prior to
6364 Note that you cannot use this option to access unaligned word objects,
6365 since the processor will only fetch one 32-bit aligned object from
6368 The default setting for most targets is @option{-mno-alignment-traps}, since
6369 this produces better code when there are no half-word memory
6370 instructions available.
6372 @item -mshort-load-bytes
6373 @itemx -mno-short-load-words
6374 @opindex mshort-load-bytes
6375 @opindex mno-short-load-words
6376 These are deprecated aliases for @option{-malignment-traps}.
6378 @item -mno-short-load-bytes
6379 @itemx -mshort-load-words
6380 @opindex mno-short-load-bytes
6381 @opindex mshort-load-words
6382 This are deprecated aliases for @option{-mno-alignment-traps}.
6384 @item -mcpu=@var{name}
6386 This specifies the name of the target ARM processor. GCC uses this name
6387 to determine what kind of instructions it can emit when generating
6388 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6389 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6390 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6391 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6392 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6393 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6394 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6395 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6396 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6397 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6399 @itemx -mtune=@var{name}
6401 This option is very similar to the @option{-mcpu=} option, except that
6402 instead of specifying the actual target processor type, and hence
6403 restricting which instructions can be used, it specifies that GCC should
6404 tune the performance of the code as if the target were of the type
6405 specified in this option, but still choosing the instructions that it
6406 will generate based on the cpu specified by a @option{-mcpu=} option.
6407 For some ARM implementations better performance can be obtained by using
6410 @item -march=@var{name}
6412 This specifies the name of the target ARM architecture. GCC uses this
6413 name to determine what kind of instructions it can emit when generating
6414 assembly code. This option can be used in conjunction with or instead
6415 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6416 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6417 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6419 @item -mfpe=@var{number}
6420 @itemx -mfp=@var{number}
6423 This specifies the version of the floating point emulation available on
6424 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6425 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6427 @item -mstructure-size-boundary=@var{n}
6428 @opindex mstructure-size-boundary
6429 The size of all structures and unions will be rounded up to a multiple
6430 of the number of bits set by this option. Permissible values are 8 and
6431 32. The default value varies for different toolchains. For the COFF
6432 targeted toolchain the default value is 8. Specifying the larger number
6433 can produce faster, more efficient code, but can also increase the size
6434 of the program. The two values are potentially incompatible. Code
6435 compiled with one value cannot necessarily expect to work with code or
6436 libraries compiled with the other value, if they exchange information
6437 using structures or unions.
6439 @item -mabort-on-noreturn
6440 @opindex mabort-on-noreturn
6441 Generate a call to the function @code{abort} at the end of a
6442 @code{noreturn} function. It will be executed if the function tries to
6446 @itemx -mno-long-calls
6447 @opindex mlong-calls
6448 @opindex mno-long-calls
6449 Tells the compiler to perform function calls by first loading the
6450 address of the function into a register and then performing a subroutine
6451 call on this register. This switch is needed if the target function
6452 will lie outside of the 64 megabyte addressing range of the offset based
6453 version of subroutine call instruction.
6455 Even if this switch is enabled, not all function calls will be turned
6456 into long calls. The heuristic is that static functions, functions
6457 which have the @samp{short-call} attribute, functions that are inside
6458 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6459 definitions have already been compiled within the current compilation
6460 unit, will not be turned into long calls. The exception to this rule is
6461 that weak function definitions, functions with the @samp{long-call}
6462 attribute or the @samp{section} attribute, and functions that are within
6463 the scope of a @samp{#pragma long_calls} directive, will always be
6464 turned into long calls.
6466 This feature is not enabled by default. Specifying
6467 @option{-mno-long-calls} will restore the default behavior, as will
6468 placing the function calls within the scope of a @samp{#pragma
6469 long_calls_off} directive. Note these switches have no effect on how
6470 the compiler generates code to handle function calls via function
6473 @item -mnop-fun-dllimport
6474 @opindex mnop-fun-dllimport
6475 Disable support for the @code{dllimport} attribute.
6477 @item -msingle-pic-base
6478 @opindex msingle-pic-base
6479 Treat the register used for PIC addressing as read-only, rather than
6480 loading it in the prologue for each function. The run-time system is
6481 responsible for initializing this register with an appropriate value
6482 before execution begins.
6484 @item -mpic-register=@var{reg}
6485 @opindex mpic-register
6486 Specify the register to be used for PIC addressing. The default is R10
6487 unless stack-checking is enabled, when R9 is used.
6489 @item -mcirrus-fix-invalid-insns
6490 @opindex mcirrus-fix-invalid-insns
6491 @opindex mno-cirrus-fix-invalid-insns
6492 Insert NOPs into the instruction stream to in order to work around
6493 problems with invalid Maverick instruction combinations. This option
6494 is only valid if the @option{-mcpu=ep9312} option has been used to
6495 enable generation of instructions for the Cirrus Maverick floating
6496 point co-processor. This option is not enabled by default, since the
6497 problem is only present in older Maverick implementations. The default
6498 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6501 @item -mpoke-function-name
6502 @opindex mpoke-function-name
6503 Write the name of each function into the text section, directly
6504 preceding the function prologue. The generated code is similar to this:
6508 .ascii "arm_poke_function_name", 0
6511 .word 0xff000000 + (t1 - t0)
6512 arm_poke_function_name
6514 stmfd sp!, @{fp, ip, lr, pc@}
6518 When performing a stack backtrace, code can inspect the value of
6519 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6520 location @code{pc - 12} and the top 8 bits are set, then we know that
6521 there is a function name embedded immediately preceding this location
6522 and has length @code{((pc[-3]) & 0xff000000)}.
6526 Generate code for the 16-bit Thumb instruction set. The default is to
6527 use the 32-bit ARM instruction set.
6530 @opindex mtpcs-frame
6531 Generate a stack frame that is compliant with the Thumb Procedure Call
6532 Standard for all non-leaf functions. (A leaf function is one that does
6533 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6535 @item -mtpcs-leaf-frame
6536 @opindex mtpcs-leaf-frame
6537 Generate a stack frame that is compliant with the Thumb Procedure Call
6538 Standard for all leaf functions. (A leaf function is one that does
6539 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6541 @item -mcallee-super-interworking
6542 @opindex mcallee-super-interworking
6543 Gives all externally visible functions in the file being compiled an ARM
6544 instruction set header which switches to Thumb mode before executing the
6545 rest of the function. This allows these functions to be called from
6546 non-interworking code.
6548 @item -mcaller-super-interworking
6549 @opindex mcaller-super-interworking
6550 Allows calls via function pointers (including virtual functions) to
6551 execute correctly regardless of whether the target code has been
6552 compiled for interworking or not. There is a small overhead in the cost
6553 of executing a function pointer if this option is enabled.
6557 @node MN10200 Options
6558 @subsection MN10200 Options
6559 @cindex MN10200 options
6561 These @option{-m} options are defined for Matsushita MN10200 architectures:
6566 Indicate to the linker that it should perform a relaxation optimization pass
6567 to shorten branches, calls and absolute memory addresses. This option only
6568 has an effect when used on the command line for the final link step.
6570 This option makes symbolic debugging impossible.
6573 @node MN10300 Options
6574 @subsection MN10300 Options
6575 @cindex MN10300 options
6577 These @option{-m} options are defined for Matsushita MN10300 architectures:
6582 Generate code to avoid bugs in the multiply instructions for the MN10300
6583 processors. This is the default.
6586 @opindex mno-mult-bug
6587 Do not generate code to avoid bugs in the multiply instructions for the
6592 Generate code which uses features specific to the AM33 processor.
6596 Do not generate code which uses features specific to the AM33 processor. This
6601 Do not link in the C run-time initialization object file.
6605 Indicate to the linker that it should perform a relaxation optimization pass
6606 to shorten branches, calls and absolute memory addresses. This option only
6607 has an effect when used on the command line for the final link step.
6609 This option makes symbolic debugging impossible.
6613 @node M32R/D Options
6614 @subsection M32R/D Options
6615 @cindex M32R/D options
6617 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6622 Generate code for the M32R/X@.
6626 Generate code for the M32R@. This is the default.
6628 @item -mcode-model=small
6629 @opindex mcode-model=small
6630 Assume all objects live in the lower 16MB of memory (so that their addresses
6631 can be loaded with the @code{ld24} instruction), and assume all subroutines
6632 are reachable with the @code{bl} instruction.
6633 This is the default.
6635 The addressability of a particular object can be set with the
6636 @code{model} attribute.
6638 @item -mcode-model=medium
6639 @opindex mcode-model=medium
6640 Assume objects may be anywhere in the 32-bit address space (the compiler
6641 will generate @code{seth/add3} instructions to load their addresses), and
6642 assume all subroutines are reachable with the @code{bl} instruction.
6644 @item -mcode-model=large
6645 @opindex mcode-model=large
6646 Assume objects may be anywhere in the 32-bit address space (the compiler
6647 will generate @code{seth/add3} instructions to load their addresses), and
6648 assume subroutines may not be reachable with the @code{bl} instruction
6649 (the compiler will generate the much slower @code{seth/add3/jl}
6650 instruction sequence).
6653 @opindex msdata=none
6654 Disable use of the small data area. Variables will be put into
6655 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6656 @code{section} attribute has been specified).
6657 This is the default.
6659 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6660 Objects may be explicitly put in the small data area with the
6661 @code{section} attribute using one of these sections.
6664 @opindex msdata=sdata
6665 Put small global and static data in the small data area, but do not
6666 generate special code to reference them.
6670 Put small global and static data in the small data area, and generate
6671 special instructions to reference them.
6675 @cindex smaller data references
6676 Put global and static objects less than or equal to @var{num} bytes
6677 into the small data or bss sections instead of the normal data or bss
6678 sections. The default value of @var{num} is 8.
6679 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6680 for this option to have any effect.
6682 All modules should be compiled with the same @option{-G @var{num}} value.
6683 Compiling with different values of @var{num} may or may not work; if it
6684 doesn't the linker will give an error message---incorrect code will not be
6690 @subsection M88K Options
6691 @cindex M88k options
6693 These @samp{-m} options are defined for Motorola 88k architectures:
6698 Generate code that works well on both the m88100 and the
6703 Generate code that works best for the m88100, but that also
6708 Generate code that works best for the m88110, and may not run
6713 Obsolete option to be removed from the next revision.
6716 @item -midentify-revision
6717 @opindex midentify-revision
6718 @cindex identifying source, compiler (88k)
6719 Include an @code{ident} directive in the assembler output recording the
6720 source file name, compiler name and version, timestamp, and compilation
6723 @item -mno-underscores
6724 @opindex mno-underscores
6725 @cindex underscores, avoiding (88k)
6726 In assembler output, emit symbol names without adding an underscore
6727 character at the beginning of each name. The default is to use an
6728 underscore as prefix on each name.
6730 @item -mocs-debug-info
6731 @itemx -mno-ocs-debug-info
6732 @opindex mocs-debug-info
6733 @opindex mno-ocs-debug-info
6735 @cindex debugging, 88k OCS
6736 Include (or omit) additional debugging information (about registers used
6737 in each stack frame) as specified in the 88open Object Compatibility
6738 Standard, ``OCS''@. This extra information allows debugging of code that
6739 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6740 SVr3.2 is to include this information; other 88k configurations omit this
6741 information by default.
6743 @item -mocs-frame-position
6744 @opindex mocs-frame-position
6745 @cindex register positions in frame (88k)
6746 When emitting COFF debugging information for automatic variables and
6747 parameters stored on the stack, use the offset from the canonical frame
6748 address, which is the stack pointer (register 31) on entry to the
6749 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6750 @option{-mocs-frame-position}; other 88k configurations have the default
6751 @option{-mno-ocs-frame-position}.
6753 @item -mno-ocs-frame-position
6754 @opindex mno-ocs-frame-position
6755 @cindex register positions in frame (88k)
6756 When emitting COFF debugging information for automatic variables and
6757 parameters stored on the stack, use the offset from the frame pointer
6758 register (register 30). When this option is in effect, the frame
6759 pointer is not eliminated when debugging information is selected by the
6762 @item -moptimize-arg-area
6763 @opindex moptimize-arg-area
6764 @cindex arguments in frame (88k)
6765 Save space by reorganizing the stack frame. This option generates code
6766 that does not agree with the 88open specifications, but uses less
6769 @itemx -mno-optimize-arg-area
6770 @opindex mno-optimize-arg-area
6771 Do not reorganize the stack frame to save space. This is the default.
6772 The generated conforms to the specification, but uses more memory.
6774 @item -mshort-data-@var{num}
6775 @opindex mshort-data
6776 @cindex smaller data references (88k)
6777 @cindex r0-relative references (88k)
6778 Generate smaller data references by making them relative to @code{r0},
6779 which allows loading a value using a single instruction (rather than the
6780 usual two). You control which data references are affected by
6781 specifying @var{num} with this option. For example, if you specify
6782 @option{-mshort-data-512}, then the data references affected are those
6783 involving displacements of less than 512 bytes.
6784 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6787 @item -mserialize-volatile
6788 @opindex mserialize-volatile
6789 @itemx -mno-serialize-volatile
6790 @opindex mno-serialize-volatile
6791 @cindex sequential consistency on 88k
6792 Do, or don't, generate code to guarantee sequential consistency
6793 of volatile memory references. By default, consistency is
6796 The order of memory references made by the MC88110 processor does
6797 not always match the order of the instructions requesting those
6798 references. In particular, a load instruction may execute before
6799 a preceding store instruction. Such reordering violates
6800 sequential consistency of volatile memory references, when there
6801 are multiple processors. When consistency must be guaranteed,
6802 GCC generates special instructions, as needed, to force
6803 execution in the proper order.
6805 The MC88100 processor does not reorder memory references and so
6806 always provides sequential consistency. However, by default, GCC
6807 generates the special instructions to guarantee consistency
6808 even when you use @option{-m88100}, so that the code may be run on an
6809 MC88110 processor. If you intend to run your code only on the
6810 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6812 The extra code generated to guarantee consistency may affect the
6813 performance of your application. If you know that you can safely
6814 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6820 @cindex assembler syntax, 88k
6822 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6823 related to System V release 4 (SVr4). This controls the following:
6827 Which variant of the assembler syntax to emit.
6829 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6830 that is used on System V release 4.
6832 @option{-msvr4} makes GCC issue additional declaration directives used in
6836 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6837 @option{-msvr3} is the default for all other m88k configurations.
6839 @item -mversion-03.00
6840 @opindex mversion-03.00
6841 This option is obsolete, and is ignored.
6842 @c ??? which asm syntax better for GAS? option there too?
6844 @item -mno-check-zero-division
6845 @itemx -mcheck-zero-division
6846 @opindex mno-check-zero-division
6847 @opindex mcheck-zero-division
6848 @cindex zero division on 88k
6849 Do, or don't, generate code to guarantee that integer division by
6850 zero will be detected. By default, detection is guaranteed.
6852 Some models of the MC88100 processor fail to trap upon integer
6853 division by zero under certain conditions. By default, when
6854 compiling code that might be run on such a processor, GCC
6855 generates code that explicitly checks for zero-valued divisors
6856 and traps with exception number 503 when one is detected. Use of
6857 @option{-mno-check-zero-division} suppresses such checking for code
6858 generated to run on an MC88100 processor.
6860 GCC assumes that the MC88110 processor correctly detects all instances
6861 of integer division by zero. When @option{-m88110} is specified, no
6862 explicit checks for zero-valued divisors are generated, and both
6863 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6866 @item -muse-div-instruction
6867 @opindex muse-div-instruction
6868 @cindex divide instruction, 88k
6869 Use the div instruction for signed integer division on the
6870 MC88100 processor. By default, the div instruction is not used.
6872 On the MC88100 processor the signed integer division instruction
6873 div) traps to the operating system on a negative operand. The
6874 operating system transparently completes the operation, but at a
6875 large cost in execution time. By default, when compiling code
6876 that might be run on an MC88100 processor, GCC emulates signed
6877 integer division using the unsigned integer division instruction
6878 divu), thereby avoiding the large penalty of a trap to the
6879 operating system. Such emulation has its own, smaller, execution
6880 cost in both time and space. To the extent that your code's
6881 important signed integer division operations are performed on two
6882 nonnegative operands, it may be desirable to use the div
6883 instruction directly.
6885 On the MC88110 processor the div instruction (also known as the
6886 divs instruction) processes negative operands without trapping to
6887 the operating system. When @option{-m88110} is specified,
6888 @option{-muse-div-instruction} is ignored, and the div instruction is used
6889 for signed integer division.
6891 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6892 particular, the behavior of such a division with and without
6893 @option{-muse-div-instruction} may differ.
6895 @item -mtrap-large-shift
6896 @itemx -mhandle-large-shift
6897 @opindex mtrap-large-shift
6898 @opindex mhandle-large-shift
6899 @cindex bit shift overflow (88k)
6900 @cindex large bit shifts (88k)
6901 Include code to detect bit-shifts of more than 31 bits; respectively,
6902 trap such shifts or emit code to handle them properly. By default GCC
6903 makes no special provision for large bit shifts.
6905 @item -mwarn-passed-structs
6906 @opindex mwarn-passed-structs
6907 @cindex structure passing (88k)
6908 Warn when a function passes a struct as an argument or result.
6909 Structure-passing conventions have changed during the evolution of the C
6910 language, and are often the source of portability problems. By default,
6911 GCC issues no such warning.
6914 @c break page here to avoid unsightly interparagraph stretch.
6918 @node RS/6000 and PowerPC Options
6919 @subsection IBM RS/6000 and PowerPC Options
6920 @cindex RS/6000 and PowerPC Options
6921 @cindex IBM RS/6000 and PowerPC Options
6923 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6931 @itemx -mpowerpc-gpopt
6932 @itemx -mno-powerpc-gpopt
6933 @itemx -mpowerpc-gfxopt
6934 @itemx -mno-powerpc-gfxopt
6936 @itemx -mno-powerpc64
6942 @opindex mno-powerpc
6943 @opindex mpowerpc-gpopt
6944 @opindex mno-powerpc-gpopt
6945 @opindex mpowerpc-gfxopt
6946 @opindex mno-powerpc-gfxopt
6948 @opindex mno-powerpc64
6949 GCC supports two related instruction set architectures for the
6950 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6951 instructions supported by the @samp{rios} chip set used in the original
6952 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6953 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6954 the IBM 4xx microprocessors.
6956 Neither architecture is a subset of the other. However there is a
6957 large common subset of instructions supported by both. An MQ
6958 register is included in processors supporting the POWER architecture.
6960 You use these options to specify which instructions are available on the
6961 processor you are using. The default value of these options is
6962 determined when configuring GCC@. Specifying the
6963 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6964 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6965 rather than the options listed above.
6967 The @option{-mpower} option allows GCC to generate instructions that
6968 are found only in the POWER architecture and to use the MQ register.
6969 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6970 to generate instructions that are present in the POWER2 architecture but
6971 not the original POWER architecture.
6973 The @option{-mpowerpc} option allows GCC to generate instructions that
6974 are found only in the 32-bit subset of the PowerPC architecture.
6975 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6976 GCC to use the optional PowerPC architecture instructions in the
6977 General Purpose group, including floating-point square root. Specifying
6978 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6979 use the optional PowerPC architecture instructions in the Graphics
6980 group, including floating-point select.
6982 The @option{-mpowerpc64} option allows GCC to generate the additional
6983 64-bit instructions that are found in the full PowerPC64 architecture
6984 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6985 @option{-mno-powerpc64}.
6987 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6988 will use only the instructions in the common subset of both
6989 architectures plus some special AIX common-mode calls, and will not use
6990 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6991 permits GCC to use any instruction from either architecture and to
6992 allow use of the MQ register; specify this for the Motorola MPC601.
6994 @item -mnew-mnemonics
6995 @itemx -mold-mnemonics
6996 @opindex mnew-mnemonics
6997 @opindex mold-mnemonics
6998 Select which mnemonics to use in the generated assembler code. With
6999 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7000 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7001 assembler mnemonics defined for the POWER architecture. Instructions
7002 defined in only one architecture have only one mnemonic; GCC uses that
7003 mnemonic irrespective of which of these options is specified.
7005 GCC defaults to the mnemonics appropriate for the architecture in
7006 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7007 value of these option. Unless you are building a cross-compiler, you
7008 should normally not specify either @option{-mnew-mnemonics} or
7009 @option{-mold-mnemonics}, but should instead accept the default.
7011 @item -mcpu=@var{cpu_type}
7013 Set architecture type, register usage, choice of mnemonics, and
7014 instruction scheduling parameters for machine type @var{cpu_type}.
7015 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7016 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7017 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7018 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7019 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7020 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7022 @option{-mcpu=common} selects a completely generic processor. Code
7023 generated under this option will run on any POWER or PowerPC processor.
7024 GCC will use only the instructions in the common subset of both
7025 architectures, and will not use the MQ register. GCC assumes a generic
7026 processor model for scheduling purposes.
7028 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7029 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7030 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7031 types, with an appropriate, generic processor model assumed for
7032 scheduling purposes.
7034 The other options specify a specific processor. Code generated under
7035 those options will run best on that processor, and may not run at all on
7038 The @option{-mcpu} options automatically enable or disable other
7039 @option{-m} options as follows:
7043 @option{-mno-power}, @option{-mno-powerpc}
7050 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7065 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7068 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7073 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7076 @item -mtune=@var{cpu_type}
7078 Set the instruction scheduling parameters for machine type
7079 @var{cpu_type}, but do not set the architecture type, register usage, or
7080 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7081 values for @var{cpu_type} are used for @option{-mtune} as for
7082 @option{-mcpu}. If both are specified, the code generated will use the
7083 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7084 scheduling parameters set by @option{-mtune}.
7089 @opindex mno-altivec
7090 These switches enable or disable the use of built-in functions that
7091 allow access to the AltiVec instruction set. You may also need to set
7092 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7097 Extend the current ABI with SPE ABI extensions. This does not change
7098 the default ABI, instead it adds the SPE ABI extensions to the current
7102 @opindex mabi=no-spe
7103 Disable Booke SPE ABI extensions for the current ABI.
7105 @item -misel=@var{yes/no}
7108 This switch enables or disables the generation of ISEL instructions.
7110 @item -mspe=@var{yes/no}
7113 This switch enables or disables the generation of SPE simd
7116 @item -mfloat-gprs=@var{yes/no}
7118 @opindex mfloat-gprs
7119 This switch enables or disables the generation of floating point
7120 operations on the general purpose registers for architectures that
7121 support it. This option is currently only available on the MPC8540.
7124 @itemx -mno-fp-in-toc
7125 @itemx -mno-sum-in-toc
7126 @itemx -mminimal-toc
7128 @opindex mno-fp-in-toc
7129 @opindex mno-sum-in-toc
7130 @opindex mminimal-toc
7131 Modify generation of the TOC (Table Of Contents), which is created for
7132 every executable file. The @option{-mfull-toc} option is selected by
7133 default. In that case, GCC will allocate at least one TOC entry for
7134 each unique non-automatic variable reference in your program. GCC
7135 will also place floating-point constants in the TOC@. However, only
7136 16,384 entries are available in the TOC@.
7138 If you receive a linker error message that saying you have overflowed
7139 the available TOC space, you can reduce the amount of TOC space used
7140 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7141 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7142 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7143 generate code to calculate the sum of an address and a constant at
7144 run-time instead of putting that sum into the TOC@. You may specify one
7145 or both of these options. Each causes GCC to produce very slightly
7146 slower and larger code at the expense of conserving TOC space.
7148 If you still run out of space in the TOC even when you specify both of
7149 these options, specify @option{-mminimal-toc} instead. This option causes
7150 GCC to make only one TOC entry for every file. When you specify this
7151 option, GCC will produce code that is slower and larger but which
7152 uses extremely little TOC space. You may wish to use this option
7153 only on files that contain less frequently executed code.
7159 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7160 @code{long} type, and the infrastructure needed to support them.
7161 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7162 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7163 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7168 @opindex mno-xl-call
7169 On AIX, pass floating-point arguments to prototyped functions beyond the
7170 register save area (RSA) on the stack in addition to argument FPRs. The
7171 AIX calling convention was extended but not initially documented to
7172 handle an obscure K&R C case of calling a function that takes the
7173 address of its arguments with fewer arguments than declared. AIX XL
7174 compilers access floating point arguments which do not fit in the
7175 RSA from the stack when a subroutine is compiled without
7176 optimization. Because always storing floating-point arguments on the
7177 stack is inefficient and rarely needed, this option is not enabled by
7178 default and only is necessary when calling subroutines compiled by AIX
7179 XL compilers without optimization.
7183 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7184 application written to use message passing with special startup code to
7185 enable the application to run. The system must have PE installed in the
7186 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7187 must be overridden with the @option{-specs=} option to specify the
7188 appropriate directory location. The Parallel Environment does not
7189 support threads, so the @option{-mpe} option and the @option{-pthread}
7190 option are incompatible.
7192 @item -malign-natural
7193 @itemx -malign-power
7194 @opindex malign-natural
7195 @opindex malign-power
7196 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7197 @option{-malign-natural} overrides the ABI-defined alignment of larger
7198 types, such as floating-point doubles, on their natural size-based boundary.
7199 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7200 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7204 @opindex msoft-float
7205 @opindex mhard-float
7206 Generate code that does not use (uses) the floating-point register set.
7207 Software floating point emulation is provided if you use the
7208 @option{-msoft-float} option, and pass the option to GCC when linking.
7211 @itemx -mno-multiple
7213 @opindex mno-multiple
7214 Generate code that uses (does not use) the load multiple word
7215 instructions and the store multiple word instructions. These
7216 instructions are generated by default on POWER systems, and not
7217 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7218 endian PowerPC systems, since those instructions do not work when the
7219 processor is in little endian mode. The exceptions are PPC740 and
7220 PPC750 which permit the instructions usage in little endian mode.
7226 Generate code that uses (does not use) the load string instructions
7227 and the store string word instructions to save multiple registers and
7228 do small block moves. These instructions are generated by default on
7229 POWER systems, and not generated on PowerPC systems. Do not use
7230 @option{-mstring} on little endian PowerPC systems, since those
7231 instructions do not work when the processor is in little endian mode.
7232 The exceptions are PPC740 and PPC750 which permit the instructions
7233 usage in little endian mode.
7239 Generate code that uses (does not use) the load or store instructions
7240 that update the base register to the address of the calculated memory
7241 location. These instructions are generated by default. If you use
7242 @option{-mno-update}, there is a small window between the time that the
7243 stack pointer is updated and the address of the previous frame is
7244 stored, which means code that walks the stack frame across interrupts or
7245 signals may get corrupted data.
7248 @itemx -mno-fused-madd
7249 @opindex mfused-madd
7250 @opindex mno-fused-madd
7251 Generate code that uses (does not use) the floating point multiply and
7252 accumulate instructions. These instructions are generated by default if
7253 hardware floating is used.
7255 @item -mno-bit-align
7257 @opindex mno-bit-align
7259 On System V.4 and embedded PowerPC systems do not (do) force structures
7260 and unions that contain bit-fields to be aligned to the base type of the
7263 For example, by default a structure containing nothing but 8
7264 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7265 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7266 the structure would be aligned to a 1 byte boundary and be one byte in
7269 @item -mno-strict-align
7270 @itemx -mstrict-align
7271 @opindex mno-strict-align
7272 @opindex mstrict-align
7273 On System V.4 and embedded PowerPC systems do not (do) assume that
7274 unaligned memory references will be handled by the system.
7277 @itemx -mno-relocatable
7278 @opindex mrelocatable
7279 @opindex mno-relocatable
7280 On embedded PowerPC systems generate code that allows (does not allow)
7281 the program to be relocated to a different address at runtime. If you
7282 use @option{-mrelocatable} on any module, all objects linked together must
7283 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7285 @item -mrelocatable-lib
7286 @itemx -mno-relocatable-lib
7287 @opindex mrelocatable-lib
7288 @opindex mno-relocatable-lib
7289 On embedded PowerPC systems generate code that allows (does not allow)
7290 the program to be relocated to a different address at runtime. Modules
7291 compiled with @option{-mrelocatable-lib} can be linked with either modules
7292 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7293 with modules compiled with the @option{-mrelocatable} options.
7299 On System V.4 and embedded PowerPC systems do not (do) assume that
7300 register 2 contains a pointer to a global area pointing to the addresses
7301 used in the program.
7304 @itemx -mlittle-endian
7306 @opindex mlittle-endian
7307 On System V.4 and embedded PowerPC systems compile code for the
7308 processor in little endian mode. The @option{-mlittle-endian} option is
7309 the same as @option{-mlittle}.
7314 @opindex mbig-endian
7315 On System V.4 and embedded PowerPC systems compile code for the
7316 processor in big endian mode. The @option{-mbig-endian} option is
7317 the same as @option{-mbig}.
7319 @item -mdynamic-no-pic
7320 @opindex mdynamic-no-pic
7321 On Darwin and Mac OS X systems, compile code so that it is not
7322 relocatable, but that its external references are relocatable. The
7323 resulting code is suitable for applications, but not shared
7328 On System V.4 and embedded PowerPC systems compile code using calling
7329 conventions that adheres to the March 1995 draft of the System V
7330 Application Binary Interface, PowerPC processor supplement. This is the
7331 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7333 @item -mcall-sysv-eabi
7334 @opindex mcall-sysv-eabi
7335 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7337 @item -mcall-sysv-noeabi
7338 @opindex mcall-sysv-noeabi
7339 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7341 @item -mcall-solaris
7342 @opindex mcall-solaris
7343 On System V.4 and embedded PowerPC systems compile code for the Solaris
7347 @opindex mcall-linux
7348 On System V.4 and embedded PowerPC systems compile code for the
7349 Linux-based GNU system.
7353 On System V.4 and embedded PowerPC systems compile code for the
7354 Hurd-based GNU system.
7357 @opindex mcall-netbsd
7358 On System V.4 and embedded PowerPC systems compile code for the
7359 NetBSD operating system.
7361 @item -maix-struct-return
7362 @opindex maix-struct-return
7363 Return all structures in memory (as specified by the AIX ABI)@.
7365 @item -msvr4-struct-return
7366 @opindex msvr4-struct-return
7367 Return structures smaller than 8 bytes in registers (as specified by the
7371 @opindex mabi=altivec
7372 Extend the current ABI with AltiVec ABI extensions. This does not
7373 change the default ABI, instead it adds the AltiVec ABI extensions to
7376 @item -mabi=no-altivec
7377 @opindex mabi=no-altivec
7378 Disable AltiVec ABI extensions for the current ABI.
7381 @itemx -mno-prototype
7383 @opindex mno-prototype
7384 On System V.4 and embedded PowerPC systems assume that all calls to
7385 variable argument functions are properly prototyped. Otherwise, the
7386 compiler must insert an instruction before every non prototyped call to
7387 set or clear bit 6 of the condition code register (@var{CR}) to
7388 indicate whether floating point values were passed in the floating point
7389 registers in case the function takes a variable arguments. With
7390 @option{-mprototype}, only calls to prototyped variable argument functions
7391 will set or clear the bit.
7395 On embedded PowerPC systems, assume that the startup module is called
7396 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7397 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7402 On embedded PowerPC systems, assume that the startup module is called
7403 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7408 On embedded PowerPC systems, assume that the startup module is called
7409 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7413 @opindex myellowknife
7414 On embedded PowerPC systems, assume that the startup module is called
7415 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7420 On System V.4 and embedded PowerPC systems, specify that you are
7421 compiling for a VxWorks system.
7425 Specify that you are compiling for the WindISS simulation environment.
7429 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7430 header to indicate that @samp{eabi} extended relocations are used.
7436 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7437 Embedded Applications Binary Interface (eabi) which is a set of
7438 modifications to the System V.4 specifications. Selecting @option{-meabi}
7439 means that the stack is aligned to an 8 byte boundary, a function
7440 @code{__eabi} is called to from @code{main} to set up the eabi
7441 environment, and the @option{-msdata} option can use both @code{r2} and
7442 @code{r13} to point to two separate small data areas. Selecting
7443 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7444 do not call an initialization function from @code{main}, and the
7445 @option{-msdata} option will only use @code{r13} to point to a single
7446 small data area. The @option{-meabi} option is on by default if you
7447 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7450 @opindex msdata=eabi
7451 On System V.4 and embedded PowerPC systems, put small initialized
7452 @code{const} global and static data in the @samp{.sdata2} section, which
7453 is pointed to by register @code{r2}. Put small initialized
7454 non-@code{const} global and static data in the @samp{.sdata} section,
7455 which is pointed to by register @code{r13}. Put small uninitialized
7456 global and static data in the @samp{.sbss} section, which is adjacent to
7457 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7458 incompatible with the @option{-mrelocatable} option. The
7459 @option{-msdata=eabi} option also sets the @option{-memb} option.
7462 @opindex msdata=sysv
7463 On System V.4 and embedded PowerPC systems, put small global and static
7464 data in the @samp{.sdata} section, which is pointed to by register
7465 @code{r13}. Put small uninitialized global and static data in the
7466 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7467 The @option{-msdata=sysv} option is incompatible with the
7468 @option{-mrelocatable} option.
7470 @item -msdata=default
7472 @opindex msdata=default
7474 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7475 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7476 same as @option{-msdata=sysv}.
7479 @opindex msdata-data
7480 On System V.4 and embedded PowerPC systems, put small global and static
7481 data in the @samp{.sdata} section. Put small uninitialized global and
7482 static data in the @samp{.sbss} section. Do not use register @code{r13}
7483 to address small data however. This is the default behavior unless
7484 other @option{-msdata} options are used.
7488 @opindex msdata=none
7490 On embedded PowerPC systems, put all initialized global and static data
7491 in the @samp{.data} section, and all uninitialized data in the
7492 @samp{.bss} section.
7496 @cindex smaller data references (PowerPC)
7497 @cindex .sdata/.sdata2 references (PowerPC)
7498 On embedded PowerPC systems, put global and static items less than or
7499 equal to @var{num} bytes into the small data or bss sections instead of
7500 the normal data or bss section. By default, @var{num} is 8. The
7501 @option{-G @var{num}} switch is also passed to the linker.
7502 All modules should be compiled with the same @option{-G @var{num}} value.
7505 @itemx -mno-regnames
7507 @opindex mno-regnames
7508 On System V.4 and embedded PowerPC systems do (do not) emit register
7509 names in the assembly language output using symbolic forms.
7512 @itemx -mno-longcall
7514 @opindex mno-longcall
7515 Default to making all function calls via pointers, so that functions
7516 which reside further than 64 megabytes (67,108,864 bytes) from the
7517 current location can be called. This setting can be overridden by the
7518 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7520 Some linkers are capable of detecting out-of-range calls and generating
7521 glue code on the fly. On these systems, long calls are unnecessary and
7522 generate slower code. As of this writing, the AIX linker can do this,
7523 as can the GNU linker for PowerPC/64. It is planned to add this feature
7524 to the GNU linker for 32-bit PowerPC systems as well.
7526 In the future, we may cause GCC to ignore all longcall specifications
7527 when the linker is known to generate glue.
7531 Adds support for multithreading with the @dfn{pthreads} library.
7532 This option sets flags for both the preprocessor and linker.
7536 @node Darwin Options
7537 @subsection Darwin Options
7538 @cindex Darwin options
7540 These options are defined for all architectures running the Darwin operating
7541 system. They are useful for compatibility with other Mac OS compilers.
7546 Loads all members of static archive libraries.
7547 See man ld(1) for more information.
7549 @item -arch_errors_fatal
7550 @opindex arch_errors_fatal
7551 Cause the errors having to do with files that have the wrong architecture
7555 @opindex bind_at_load
7556 Causes the output file to be marked such that the dynamic linker will
7557 bind all undefined references when the file is loaded or launched.
7561 Produce a Mach-o bundle format file.
7562 See man ld(1) for more information.
7564 @item -bundle_loader @var{executable}
7565 @opindex bundle_loader
7566 This specifies the @var{executable} that will be loading the build
7567 output file being linked. See man ld(1) for more information.
7569 @item -allowable_client @var{client_name}
7573 @item -compatibility_version
7574 @item -current_version
7575 @item -dependency-file
7577 @item -dylinker_install_name
7580 @item -exported_symbols_list
7582 @item -flat_namespace
7583 @item -force_cpusubtype_ALL
7584 @item -force_flat_namespace
7585 @item -headerpad_max_install_names
7589 @item -keep_private_externs
7591 @item -multiply_defined
7592 @item -multiply_defined_unused
7596 @item -noseglinkedit
7597 @item -pagezero_size
7599 @item -prebind_all_twolevel_modules
7600 @item -private_bundle
7601 @item -read_only_relocs
7603 @item -sectobjectsymbols
7607 @item -sectobjectsymbols
7609 @item -seg_addr_table
7610 @item -seg_addr_table_filename
7613 @item -segs_read_only_addr
7614 @item -segs_read_write_addr
7615 @item -single_module
7619 @item -twolevel_namespace
7622 @item -unexported_symbols_list
7623 @item -weak_reference_mismatches
7626 @opindex allowable_client
7628 @opindex client_name
7629 @opindex compatibility_version
7630 @opindex current_version
7631 @opindex dependency-file
7633 @opindex dylinker_install_name
7636 @opindex exported_symbols_list
7638 @opindex flat_namespace
7639 @opindex force_cpusubtype_ALL
7640 @opindex force_flat_namespace
7641 @opindex headerpad_max_install_names
7644 @opindex install_name
7645 @opindex keep_private_externs
7646 @opindex multi_module
7647 @opindex multiply_defined
7648 @opindex multiply_defined_unused
7650 @opindex nomultidefs
7652 @opindex noseglinkedit
7653 @opindex pagezero_size
7655 @opindex prebind_all_twolevel_modules
7656 @opindex private_bundle
7657 @opindex read_only_relocs
7659 @opindex sectobjectsymbols
7663 @opindex sectobjectsymbols
7665 @opindex seg_addr_table
7666 @opindex seg_addr_table_filename
7667 @opindex seglinkedit
7669 @opindex segs_read_only_addr
7670 @opindex segs_read_write_addr
7671 @opindex single_module
7673 @opindex sub_library
7674 @opindex sub_umbrella
7675 @opindex twolevel_namespace
7678 @opindex unexported_symbols_list
7679 @opindex weak_reference_mismatches
7680 @opindex whatsloaded
7682 This options are available for Darwin linker. Darwin linker man page
7683 describes them in detail.
7688 @subsection IBM RT Options
7690 @cindex IBM RT options
7692 These @samp{-m} options are defined for the IBM RT PC:
7696 @opindex min-line-mul
7697 Use an in-line code sequence for integer multiplies. This is the
7700 @item -mcall-lib-mul
7701 @opindex mcall-lib-mul
7702 Call @code{lmul$$} for integer multiples.
7704 @item -mfull-fp-blocks
7705 @opindex mfull-fp-blocks
7706 Generate full-size floating point data blocks, including the minimum
7707 amount of scratch space recommended by IBM@. This is the default.
7709 @item -mminimum-fp-blocks
7710 @opindex mminimum-fp-blocks
7711 Do not include extra scratch space in floating point data blocks. This
7712 results in smaller code, but slower execution, since scratch space must
7713 be allocated dynamically.
7715 @cindex @file{stdarg.h} and RT PC
7716 @item -mfp-arg-in-fpregs
7717 @opindex mfp-arg-in-fpregs
7718 Use a calling sequence incompatible with the IBM calling convention in
7719 which floating point arguments are passed in floating point registers.
7720 Note that @code{stdarg.h} will not work with floating point operands
7721 if this option is specified.
7723 @item -mfp-arg-in-gregs
7724 @opindex mfp-arg-in-gregs
7725 Use the normal calling convention for floating point arguments. This is
7728 @item -mhc-struct-return
7729 @opindex mhc-struct-return
7730 Return structures of more than one word in memory, rather than in a
7731 register. This provides compatibility with the MetaWare HighC (hc)
7732 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7733 with the Portable C Compiler (pcc).
7735 @item -mnohc-struct-return
7736 @opindex mnohc-struct-return
7737 Return some structures of more than one word in registers, when
7738 convenient. This is the default. For compatibility with the
7739 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7740 option @option{-mhc-struct-return}.
7744 @subsection MIPS Options
7745 @cindex MIPS options
7747 These @samp{-m} options are defined for the MIPS family of computers:
7751 @item -march=@var{arch}
7753 Generate code that will run on @var{arch}, which can be the name of a
7754 generic MIPS ISA, or the name of a particular processor.
7756 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7757 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7758 The processor names are:
7759 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7761 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7762 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
7765 @samp{vr4100}, @samp{vr4300}, and @samp{vr5000}.
7766 The special value @samp{from-abi} selects the
7767 most compatible architecture for the selected ABI (that is,
7768 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7770 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7771 (for example, @samp{-march=r2k}). Prefixes are optional, and
7772 @samp{vr} may be written @samp{r}.
7774 GCC defines two macros based on the value of this option. The first
7775 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7776 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7777 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7778 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7779 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7781 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7782 above. In other words, it will have the full prefix and will not
7783 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7784 the macro names the resolved architecture (either @samp{"mips1"} or
7785 @samp{"mips3"}). It names the default architecture when no
7786 @option{-march} option is given.
7788 @item -mtune=@var{arch}
7790 Optimize for @var{arch}. Among other things, this option controls
7791 the way instructions are scheduled, and the perceived cost of arithmetic
7792 operations. The list of @var{arch} values is the same as for
7795 When this option is not used, GCC will optimize for the processor
7796 specified by @option{-march}. By using @option{-march} and
7797 @option{-mtune} together, it is possible to generate code that will
7798 run on a family of processors, but optimize the code for one
7799 particular member of that family.
7801 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7802 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7803 @samp{-march} ones described above.
7807 Equivalent to @samp{-march=mips1}.
7811 Equivalent to @samp{-march=mips2}.
7815 Equivalent to @samp{-march=mips3}.
7819 Equivalent to @samp{-march=mips4}.
7823 Equivalent to @samp{-march=mips32}.
7827 Equivalent to @samp{-march=mips32r2}.
7831 Equivalent to @samp{-march=mips64}.
7834 @itemx -mno-fused-madd
7835 @opindex mfused-madd
7836 @opindex mno-fused-madd
7837 Generate code that uses (does not use) the floating point multiply and
7838 accumulate instructions, when they are available. These instructions
7839 are generated by default if they are available, but this may be
7840 undesirable if the extra precision causes problems or on certain chips
7841 in the mode where denormals are rounded to zero where denormals
7842 generated by multiply and accumulate instructions cause exceptions
7847 Assume that floating point registers are 32 bits wide.
7851 Assume that floating point registers are 64 bits wide.
7855 Assume that general purpose registers are 32 bits wide.
7859 Assume that general purpose registers are 64 bits wide.
7863 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7864 explanation of the default, and the width of pointers.
7868 Force long types to be 64 bits wide. See @option{-mlong32} for an
7869 explanation of the default, and the width of pointers.
7873 Force long, int, and pointer types to be 32 bits wide.
7875 The default size of ints, longs and pointers depends on the ABI@. All
7876 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7877 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7878 are the same size as longs, or the same size as integer registers,
7879 whichever is smaller.
7893 Generate code for the given ABI@.
7895 Note that there are two embedded ABIs: @option{-mabi=eabi}
7896 selects the one defined by Cygnus while @option{-meabi=meabi}
7897 selects the one defined by MIPS@. Both these ABIs have
7898 32-bit and 64-bit variants. Normally, GCC will generate
7899 64-bit code when you select a 64-bit architecture, but you
7900 can use @option{-mgp32} to get 32-bit code instead.
7902 @item -mabi-fake-default
7903 @opindex mabi-fake-default
7904 You don't want to know what this option does. No, really. I mean
7905 it. Move on to the next option.
7907 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7908 wants the default set of options to get the root of the multilib tree,
7909 and the shared library SONAMEs without any multilib-indicating
7910 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7911 we want to default to the N32 ABI, while still being binary-compatible
7912 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7913 binary-compatible means shared libraries should have the same SONAMEs,
7914 and libraries should live in the same location. Having O32 libraries
7915 in a sub-directory named say @file{o32} is not acceptable.
7917 So we trick GCC into believing that O32 is the default ABI, except
7918 that we override the default with some internal command-line
7919 processing magic. Problem is, if we stopped at that, and you then
7920 created a multilib-aware package that used the output of @command{gcc
7921 -print-multi-lib} to decide which multilibs to build, and how, and
7922 you'd find yourself in an awkward situation when you found out that
7923 some of the options listed ended up mapping to the same multilib, and
7924 none of your libraries was actually built for the multilib that
7925 @option{-print-multi-lib} claims to be the default. So we added this
7926 option that disables the default switcher, falling back to GCC's
7927 original notion of the default library. Confused yet?
7929 For short: don't ever use this option, unless you find it in the list
7930 of additional options to be used when building for multilibs, in the
7931 output of @option{gcc -print-multi-lib}.
7935 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7936 add normal debug information. This is the default for all
7937 platforms except for the OSF/1 reference platform, using the OSF/rose
7938 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7939 switches are used, the @file{mips-tfile} program will encapsulate the
7940 stabs within MIPS ECOFF@.
7944 Generate code for the GNU assembler. This is the default on the OSF/1
7945 reference platform, using the OSF/rose object format. Also, this is
7946 the default if the configure option @option{--with-gnu-as} is used.
7948 @item -msplit-addresses
7949 @itemx -mno-split-addresses
7950 @opindex msplit-addresses
7951 @opindex mno-split-addresses
7952 Generate code to load the high and low parts of address constants separately.
7953 This allows GCC to optimize away redundant loads of the high order
7954 bits of addresses. This optimization requires GNU as and GNU ld.
7955 This optimization is enabled by default for some embedded targets where
7956 GNU as and GNU ld are standard.
7962 The @option{-mrnames} switch says to output code using the MIPS software
7963 names for the registers, instead of the hardware names (ie, @var{a0}
7964 instead of @var{$4}). The only known assembler that supports this option
7965 is the Algorithmics assembler.
7971 The @option{-mmemcpy} switch makes all block moves call the appropriate
7972 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7973 generating inline code.
7976 @itemx -mno-mips-tfile
7977 @opindex mmips-tfile
7978 @opindex mno-mips-tfile
7979 The @option{-mno-mips-tfile} switch causes the compiler not
7980 postprocess the object file with the @file{mips-tfile} program,
7981 after the MIPS assembler has generated it to add debug support. If
7982 @file{mips-tfile} is not run, then no local variables will be
7983 available to the debugger. In addition, @file{stage2} and
7984 @file{stage3} objects will have the temporary file names passed to the
7985 assembler embedded in the object file, which means the objects will
7986 not compare the same. The @option{-mno-mips-tfile} switch should only
7987 be used when there are bugs in the @file{mips-tfile} program that
7988 prevents compilation.
7991 @opindex msoft-float
7992 Generate output containing library calls for floating point.
7993 @strong{Warning:} the requisite libraries are not part of GCC@.
7994 Normally the facilities of the machine's usual C compiler are used, but
7995 this can't be done directly in cross-compilation. You must make your
7996 own arrangements to provide suitable library functions for
8000 @opindex mhard-float
8001 Generate output containing floating point instructions. This is the
8002 default if you use the unmodified sources.
8005 @itemx -mno-abicalls
8007 @opindex mno-abicalls
8008 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8009 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8010 position independent code.
8013 @itemx -mno-long-calls
8014 @opindex mlong-calls
8015 @opindex mno-long-calls
8016 Do all calls with the @samp{JALR} instruction, which requires
8017 loading up a function's address into a register before the call.
8018 You need to use this switch, if you call outside of the current
8019 512 megabyte segment to functions that are not through pointers.
8021 @item -membedded-pic
8022 @itemx -mno-embedded-pic
8023 @opindex membedded-pic
8024 @opindex mno-embedded-pic
8025 Generate PIC code suitable for some embedded systems. All calls are
8026 made using PC relative address, and all data is addressed using the $gp
8027 register. No more than 65536 bytes of global data may be used. This
8028 requires GNU as and GNU ld which do most of the work. This currently
8029 only works on targets which use ECOFF; it does not work with ELF@.
8031 @item -membedded-data
8032 @itemx -mno-embedded-data
8033 @opindex membedded-data
8034 @opindex mno-embedded-data
8035 Allocate variables to the read-only data section first if possible, then
8036 next in the small data section if possible, otherwise in data. This gives
8037 slightly slower code than the default, but reduces the amount of RAM required
8038 when executing, and thus may be preferred for some embedded systems.
8040 @item -muninit-const-in-rodata
8041 @itemx -mno-uninit-const-in-rodata
8042 @opindex muninit-const-in-rodata
8043 @opindex mno-uninit-const-in-rodata
8044 When used together with @option{-membedded-data}, it will always store uninitialized
8045 const variables in the read-only data section.
8047 @item -msingle-float
8048 @itemx -mdouble-float
8049 @opindex msingle-float
8050 @opindex mdouble-float
8051 The @option{-msingle-float} switch tells gcc to assume that the floating
8052 point coprocessor only supports single precision operations, as on the
8053 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8054 double precision operations. This is the default.
8060 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8061 as on the @samp{r4650} chip.
8065 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8066 @option{-mcpu=r4650}.
8072 Enable 16-bit instructions.
8076 Use the entry and exit pseudo ops. This option can only be used with
8081 Compile code for the processor in little endian mode.
8082 The requisite libraries are assumed to exist.
8086 Compile code for the processor in big endian mode.
8087 The requisite libraries are assumed to exist.
8091 @cindex smaller data references (MIPS)
8092 @cindex gp-relative references (MIPS)
8093 Put global and static items less than or equal to @var{num} bytes into
8094 the small data or bss sections instead of the normal data or bss
8095 section. This allows the assembler to emit one word memory reference
8096 instructions based on the global pointer (@var{gp} or @var{$28}),
8097 instead of the normal two words used. By default, @var{num} is 8 when
8098 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8099 @option{-G @var{num}} switch is also passed to the assembler and linker.
8100 All modules should be compiled with the same @option{-G @var{num}}
8105 Tell the MIPS assembler to not run its preprocessor over user
8106 assembler files (with a @samp{.s} suffix) when assembling them.
8110 Pass an option to gas which will cause nops to be inserted if
8111 the read of the destination register of an mfhi or mflo instruction
8112 occurs in the following two instructions.
8116 Do not include the default crt0.
8118 @item -mflush-func=@var{func}
8119 @itemx -mno-flush-func
8120 @opindex mflush-func
8121 Specifies the function to call to flush the I and D caches, or to not
8122 call any such function. If called, the function must take the same
8123 arguments as the common @code{_flush_func()}, that is, the address of the
8124 memory range for which the cache is being flushed, the size of the
8125 memory range, and the number 3 (to flush both caches). The default
8126 depends on the target gcc was configured for, but commonly is either
8127 @samp{_flush_func} or @samp{__cpu_flush}.
8129 @item -mbranch-likely
8130 @itemx -mno-branch-likely
8131 @opindex mbranch-likely
8132 @opindex mno-branch-likely
8133 Enable or disable use of Branch Likely instructions, regardless of the
8134 default for the selected architecture. By default, Branch Likely
8135 instructions may be generated if they are supported by the selected
8136 architecture. An exception is for the MIPS32 and MIPS64 architectures
8137 and processors which implement those architectures; for those, Branch
8138 Likely instructions will not be generated by default because the MIPS32
8139 and MIPS64 architectures specifically deprecate their use.
8142 @node i386 and x86-64 Options
8143 @subsection Intel 386 and AMD x86-64 Options
8144 @cindex i386 Options
8145 @cindex x86-64 Options
8146 @cindex Intel 386 Options
8147 @cindex AMD x86-64 Options
8149 These @samp{-m} options are defined for the i386 and x86-64 family of
8153 @item -mtune=@var{cpu-type}
8155 Tune to @var{cpu-type} everything applicable about the generated code, except
8156 for the ABI and the set of available instructions. The choices for
8157 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8158 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8159 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8160 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8161 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8164 While picking a specific @var{cpu-type} will schedule things appropriately
8165 for that particular chip, the compiler will not generate any code that
8166 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8167 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8168 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8169 AMD chips as opposed to the Intel ones.
8171 @item -march=@var{cpu-type}
8173 Generate instructions for the machine type @var{cpu-type}. The choices
8174 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8175 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8177 @item -mcpu=@var{cpu-type}
8179 A deprecated synonym for @option{-mtune}.
8188 @opindex mpentiumpro
8189 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8190 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8191 These synonyms are deprecated.
8193 @item -mfpmath=@var{unit}
8195 generate floating point arithmetics for selected unit @var{unit}. the choices
8200 Use the standard 387 floating point coprocessor present majority of chips and
8201 emulated otherwise. Code compiled with this option will run almost everywhere.
8202 The temporary results are computed in 80bit precision instead of precision
8203 specified by the type resulting in slightly different results compared to most
8204 of other chips. See @option{-ffloat-store} for more detailed description.
8206 This is the default choice for i386 compiler.
8209 Use scalar floating point instructions present in the SSE instruction set.
8210 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8211 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8212 instruction set supports only single precision arithmetics, thus the double and
8213 extended precision arithmetics is still done using 387. Later version, present
8214 only in Pentium4 and the future AMD x86-64 chips supports double precision
8217 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8218 @option{-msse2} switches to enable SSE extensions and make this option
8219 effective. For x86-64 compiler, these extensions are enabled by default.
8221 The resulting code should be considerably faster in majority of cases and avoid
8222 the numerical instability problems of 387 code, but may break some existing
8223 code that expects temporaries to be 80bit.
8225 This is the default choice for x86-64 compiler.
8228 Use all SSE extensions enabled by @option{-msse2} as well as the new
8229 SSE extensions in Prescott New Instructions. @option{-mpni} also
8230 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8231 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8235 Attempt to utilize both instruction sets at once. This effectively double the
8236 amount of available registers and on chips with separate execution units for
8237 387 and SSE the execution resources too. Use this option with care, as it is
8238 still experimental, because gcc register allocator does not model separate
8239 functional units well resulting in instable performance.
8242 @item -masm=@var{dialect}
8243 @opindex masm=@var{dialect}
8244 Output asm instructions using selected @var{dialect}. Supported choices are
8245 @samp{intel} or @samp{att} (the default one).
8250 @opindex mno-ieee-fp
8251 Control whether or not the compiler uses IEEE floating point
8252 comparisons. These handle correctly the case where the result of a
8253 comparison is unordered.
8256 @opindex msoft-float
8257 Generate output containing library calls for floating point.
8258 @strong{Warning:} the requisite libraries are not part of GCC@.
8259 Normally the facilities of the machine's usual C compiler are used, but
8260 this can't be done directly in cross-compilation. You must make your
8261 own arrangements to provide suitable library functions for
8264 On machines where a function returns floating point results in the 80387
8265 register stack, some floating point opcodes may be emitted even if
8266 @option{-msoft-float} is used.
8268 @item -mno-fp-ret-in-387
8269 @opindex mno-fp-ret-in-387
8270 Do not use the FPU registers for return values of functions.
8272 The usual calling convention has functions return values of types
8273 @code{float} and @code{double} in an FPU register, even if there
8274 is no FPU@. The idea is that the operating system should emulate
8277 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8278 in ordinary CPU registers instead.
8280 @item -mno-fancy-math-387
8281 @opindex mno-fancy-math-387
8282 Some 387 emulators do not support the @code{sin}, @code{cos} and
8283 @code{sqrt} instructions for the 387. Specify this option to avoid
8284 generating those instructions. This option is the default on FreeBSD,
8285 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8286 indicates that the target cpu will always have an FPU and so the
8287 instruction will not need emulation. As of revision 2.6.1, these
8288 instructions are not generated unless you also use the
8289 @option{-funsafe-math-optimizations} switch.
8291 @item -malign-double
8292 @itemx -mno-align-double
8293 @opindex malign-double
8294 @opindex mno-align-double
8295 Control whether GCC aligns @code{double}, @code{long double}, and
8296 @code{long long} variables on a two word boundary or a one word
8297 boundary. Aligning @code{double} variables on a two word boundary will
8298 produce code that runs somewhat faster on a @samp{Pentium} at the
8299 expense of more memory.
8301 @strong{Warning:} if you use the @option{-malign-double} switch,
8302 structures containing the above types will be aligned differently than
8303 the published application binary interface specifications for the 386
8304 and will not be binary compatible with structures in code compiled
8305 without that switch.
8307 @item -m96bit-long-double
8308 @item -m128bit-long-double
8309 @opindex m96bit-long-double
8310 @opindex m128bit-long-double
8311 These switches control the size of @code{long double} type. The i386
8312 application binary interface specifies the size to be 96 bits,
8313 so @option{-m96bit-long-double} is the default in 32 bit mode.
8315 Modern architectures (Pentium and newer) would prefer @code{long double}
8316 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8317 conforming to the ABI, this would not be possible. So specifying a
8318 @option{-m128bit-long-double} will align @code{long double}
8319 to a 16 byte boundary by padding the @code{long double} with an additional
8322 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8323 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8325 Notice that neither of these options enable any extra precision over the x87
8326 standard of 80 bits for a @code{long double}.
8328 @strong{Warning:} if you override the default value for your target ABI, the
8329 structures and arrays containing @code{long double} will change their size as
8330 well as function calling convention for function taking @code{long double}
8331 will be modified. Hence they will not be binary compatible with arrays or
8332 structures in code compiled without that switch.
8336 @itemx -mno-svr3-shlib
8337 @opindex msvr3-shlib
8338 @opindex mno-svr3-shlib
8339 Control whether GCC places uninitialized local variables into the
8340 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8341 into @code{bss}. These options are meaningful only on System V Release 3.
8345 Use a different function-calling convention, in which functions that
8346 take a fixed number of arguments return with the @code{ret} @var{num}
8347 instruction, which pops their arguments while returning. This saves one
8348 instruction in the caller since there is no need to pop the arguments
8351 You can specify that an individual function is called with this calling
8352 sequence with the function attribute @samp{stdcall}. You can also
8353 override the @option{-mrtd} option by using the function attribute
8354 @samp{cdecl}. @xref{Function Attributes}.
8356 @strong{Warning:} this calling convention is incompatible with the one
8357 normally used on Unix, so you cannot use it if you need to call
8358 libraries compiled with the Unix compiler.
8360 Also, you must provide function prototypes for all functions that
8361 take variable numbers of arguments (including @code{printf});
8362 otherwise incorrect code will be generated for calls to those
8365 In addition, seriously incorrect code will result if you call a
8366 function with too many arguments. (Normally, extra arguments are
8367 harmlessly ignored.)
8369 @item -mregparm=@var{num}
8371 Control how many registers are used to pass integer arguments. By
8372 default, no registers are used to pass arguments, and at most 3
8373 registers can be used. You can control this behavior for a specific
8374 function by using the function attribute @samp{regparm}.
8375 @xref{Function Attributes}.
8377 @strong{Warning:} if you use this switch, and
8378 @var{num} is nonzero, then you must build all modules with the same
8379 value, including any libraries. This includes the system libraries and
8382 @item -mpreferred-stack-boundary=@var{num}
8383 @opindex mpreferred-stack-boundary
8384 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8385 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8386 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8387 size (@option{-Os}), in which case the default is the minimum correct
8388 alignment (4 bytes for x86, and 8 bytes for x86-64).
8390 On Pentium and PentiumPro, @code{double} and @code{long double} values
8391 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8392 suffer significant run time performance penalties. On Pentium III, the
8393 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8394 penalties if it is not 16 byte aligned.
8396 To ensure proper alignment of this values on the stack, the stack boundary
8397 must be as aligned as that required by any value stored on the stack.
8398 Further, every function must be generated such that it keeps the stack
8399 aligned. Thus calling a function compiled with a higher preferred
8400 stack boundary from a function compiled with a lower preferred stack
8401 boundary will most likely misalign the stack. It is recommended that
8402 libraries that use callbacks always use the default setting.
8404 This extra alignment does consume extra stack space, and generally
8405 increases code size. Code that is sensitive to stack space usage, such
8406 as embedded systems and operating system kernels, may want to reduce the
8407 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8425 These switches enable or disable the use of built-in functions that allow
8426 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8428 @xref{X86 Built-in Functions}, for details of the functions enabled
8429 and disabled by these switches.
8431 To have SSE/SSE2 instructions generated automatically from floating-point
8432 code, see @option{-mfpmath=sse}.
8435 @itemx -mno-push-args
8437 @opindex mno-push-args
8438 Use PUSH operations to store outgoing parameters. This method is shorter
8439 and usually equally fast as method using SUB/MOV operations and is enabled
8440 by default. In some cases disabling it may improve performance because of
8441 improved scheduling and reduced dependencies.
8443 @item -maccumulate-outgoing-args
8444 @opindex maccumulate-outgoing-args
8445 If enabled, the maximum amount of space required for outgoing arguments will be
8446 computed in the function prologue. This is faster on most modern CPUs
8447 because of reduced dependencies, improved scheduling and reduced stack usage
8448 when preferred stack boundary is not equal to 2. The drawback is a notable
8449 increase in code size. This switch implies @option{-mno-push-args}.
8453 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8454 on thread-safe exception handling must compile and link all code with the
8455 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8456 @option{-D_MT}; when linking, it links in a special thread helper library
8457 @option{-lmingwthrd} which cleans up per thread exception handling data.
8459 @item -mno-align-stringops
8460 @opindex mno-align-stringops
8461 Do not align destination of inlined string operations. This switch reduces
8462 code size and improves performance in case the destination is already aligned,
8463 but gcc don't know about it.
8465 @item -minline-all-stringops
8466 @opindex minline-all-stringops
8467 By default GCC inlines string operations only when destination is known to be
8468 aligned at least to 4 byte boundary. This enables more inlining, increase code
8469 size, but may improve performance of code that depends on fast memcpy, strlen
8470 and memset for short lengths.
8472 @item -momit-leaf-frame-pointer
8473 @opindex momit-leaf-frame-pointer
8474 Don't keep the frame pointer in a register for leaf functions. This
8475 avoids the instructions to save, set up and restore frame pointers and
8476 makes an extra register available in leaf functions. The option
8477 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8478 which might make debugging harder.
8480 @item -mtls-direct-seg-refs
8481 @itemx -mno-tls-direct-seg-refs
8482 @opindex mtls-direct-seg-refs
8483 Controls whether TLS variables may be accessed with offsets from the
8484 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8485 or whether the thread base pointer must be added. Whether or not this
8486 is legal depends on the operating system, and whether it maps the
8487 segment to cover the entire TLS area.
8489 For systems that use GNU libc, the default is on.
8492 These @samp{-m} switches are supported in addition to the above
8493 on AMD x86-64 processors in 64-bit environments.
8500 Generate code for a 32-bit or 64-bit environment.
8501 The 32-bit environment sets int, long and pointer to 32 bits and
8502 generates code that runs on any i386 system.
8503 The 64-bit environment sets int to 32 bits and long and pointer
8504 to 64 bits and generates code for AMD's x86-64 architecture.
8507 @opindex no-red-zone
8508 Do not use a so called red zone for x86-64 code. The red zone is mandated
8509 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8510 stack pointer that will not be modified by signal or interrupt handlers
8511 and therefore can be used for temporary data without adjusting the stack
8512 pointer. The flag @option{-mno-red-zone} disables this red zone.
8514 @item -mcmodel=small
8515 @opindex mcmodel=small
8516 Generate code for the small code model: the program and its symbols must
8517 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8518 Programs can be statically or dynamically linked. This is the default
8521 @item -mcmodel=kernel
8522 @opindex mcmodel=kernel
8523 Generate code for the kernel code model. The kernel runs in the
8524 negative 2 GB of the address space.
8525 This model has to be used for Linux kernel code.
8527 @item -mcmodel=medium
8528 @opindex mcmodel=medium
8529 Generate code for the medium model: The program is linked in the lower 2
8530 GB of the address space but symbols can be located anywhere in the
8531 address space. Programs can be statically or dynamically linked, but
8532 building of shared libraries are not supported with the medium model.
8534 @item -mcmodel=large
8535 @opindex mcmodel=large
8536 Generate code for the large model: This model makes no assumptions
8537 about addresses and sizes of sections. Currently GCC does not implement
8542 @subsection HPPA Options
8543 @cindex HPPA Options
8545 These @samp{-m} options are defined for the HPPA family of computers:
8548 @item -march=@var{architecture-type}
8550 Generate code for the specified architecture. The choices for
8551 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8552 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8553 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8554 architecture option for your machine. Code compiled for lower numbered
8555 architectures will run on higher numbered architectures, but not the
8558 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8559 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8563 @itemx -mpa-risc-1-1
8564 @itemx -mpa-risc-2-0
8565 @opindex mpa-risc-1-0
8566 @opindex mpa-risc-1-1
8567 @opindex mpa-risc-2-0
8568 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8571 @opindex mbig-switch
8572 Generate code suitable for big switch tables. Use this option only if
8573 the assembler/linker complain about out of range branches within a switch
8576 @item -mjump-in-delay
8577 @opindex mjump-in-delay
8578 Fill delay slots of function calls with unconditional jump instructions
8579 by modifying the return pointer for the function call to be the target
8580 of the conditional jump.
8582 @item -mdisable-fpregs
8583 @opindex mdisable-fpregs
8584 Prevent floating point registers from being used in any manner. This is
8585 necessary for compiling kernels which perform lazy context switching of
8586 floating point registers. If you use this option and attempt to perform
8587 floating point operations, the compiler will abort.
8589 @item -mdisable-indexing
8590 @opindex mdisable-indexing
8591 Prevent the compiler from using indexing address modes. This avoids some
8592 rather obscure problems when compiling MIG generated code under MACH@.
8594 @item -mno-space-regs
8595 @opindex mno-space-regs
8596 Generate code that assumes the target has no space registers. This allows
8597 GCC to generate faster indirect calls and use unscaled index address modes.
8599 Such code is suitable for level 0 PA systems and kernels.
8601 @item -mfast-indirect-calls
8602 @opindex mfast-indirect-calls
8603 Generate code that assumes calls never cross space boundaries. This
8604 allows GCC to emit code which performs faster indirect calls.
8606 This option will not work in the presence of shared libraries or nested
8609 @item -mlong-load-store
8610 @opindex mlong-load-store
8611 Generate 3-instruction load and store sequences as sometimes required by
8612 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8615 @item -mportable-runtime
8616 @opindex mportable-runtime
8617 Use the portable calling conventions proposed by HP for ELF systems.
8621 Enable the use of assembler directives only GAS understands.
8623 @item -mschedule=@var{cpu-type}
8625 Schedule code according to the constraints for the machine type
8626 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8627 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8628 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8629 proper scheduling option for your machine. The default scheduling is
8633 @opindex mlinker-opt
8634 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8635 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8636 linkers in which they give bogus error messages when linking some programs.
8639 @opindex msoft-float
8640 Generate output containing library calls for floating point.
8641 @strong{Warning:} the requisite libraries are not available for all HPPA
8642 targets. Normally the facilities of the machine's usual C compiler are
8643 used, but this cannot be done directly in cross-compilation. You must make
8644 your own arrangements to provide suitable library functions for
8645 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8646 does provide software floating point support.
8648 @option{-msoft-float} changes the calling convention in the output file;
8649 therefore, it is only useful if you compile @emph{all} of a program with
8650 this option. In particular, you need to compile @file{libgcc.a}, the
8651 library that comes with GCC, with @option{-msoft-float} in order for
8656 Generate the predefine, @code{_SIO}, for server IO. The default is
8657 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8658 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8659 options are available under HP-UX and HI-UX.
8663 Use GNU ld specific options. This passes @option{-shared} to ld when
8664 building a shared library. It is the default when GCC is configured,
8665 explicitly or implicitly, with the GNU linker. This option does not
8666 have any affect on which ld is called, it only changes what parameters
8667 are passed to that ld. The ld that is called is determined by the
8668 @option{--with-ld} configure option, gcc's program search path, and
8669 finally by the user's @env{PATH}. The linker used by GCC can be printed
8670 using @samp{which `gcc -print-prog-name=ld`}.
8674 Use HP ld specific options. This passes @option{-b} to ld when building
8675 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8676 links. It is the default when GCC is configured, explicitly or
8677 implicitly, with the HP linker. This option does not have any affect on
8678 which ld is called, it only changes what parameters are passed to that
8679 ld. The ld that is called is determined by the @option{--with-ld}
8680 configure option, gcc's program search path, and finally by the user's
8681 @env{PATH}. The linker used by GCC can be printed using @samp{which
8682 `gcc -print-prog-name=ld`}.
8685 @opindex mno-long-calls
8686 Generate code that uses long call sequences. This ensures that a call
8687 is always able to reach linker generated stubs. The default is to generate
8688 long calls only when the distance from the call site to the beginning
8689 of the function or translation unit, as the case may be, exceeds a
8690 predefined limit set by the branch type being used. The limits for
8691 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8692 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8695 Distances are measured from the beginning of functions when using the
8696 @option{-ffunction-sections} option, or when using the @option{-mgas}
8697 and @option{-mno-portable-runtime} options together under HP-UX with
8700 It is normally not desirable to use this option as it will degrade
8701 performance. However, it may be useful in large applications,
8702 particularly when partial linking is used to build the application.
8704 The types of long calls used depends on the capabilities of the
8705 assembler and linker, and the type of code being generated. The
8706 impact on systems that support long absolute calls, and long pic
8707 symbol-difference or pc-relative calls should be relatively small.
8708 However, an indirect call is used on 32-bit ELF systems in pic code
8709 and it is quite long.
8713 Suppress the generation of link options to search libdld.sl when the
8714 @option{-static} option is specified on HP-UX 10 and later.
8718 The HP-UX implementation of setlocale in libc has a dependency on
8719 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8720 when the @option{-static} option is specified, special link options
8721 are needed to resolve this dependency.
8723 On HP-UX 10 and later, the GCC driver adds the necessary options to
8724 link with libdld.sl when the @option{-static} option is specified.
8725 This causes the resulting binary to be dynamic. On the 64-bit port,
8726 the linkers generate dynamic binaries by default in any case. The
8727 @option{-nolibdld} option can be used to prevent the GCC driver from
8728 adding these link options.
8732 Add support for multithreading with the @dfn{dce thread} library
8733 under HP-UX. This option sets flags for both the preprocessor and
8737 @node Intel 960 Options
8738 @subsection Intel 960 Options
8740 These @samp{-m} options are defined for the Intel 960 implementations:
8743 @item -m@var{cpu-type}
8751 Assume the defaults for the machine type @var{cpu-type} for some of
8752 the other options, including instruction scheduling, floating point
8753 support, and addressing modes. The choices for @var{cpu-type} are
8754 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8755 @samp{sa}, and @samp{sb}.
8762 @opindex msoft-float
8763 The @option{-mnumerics} option indicates that the processor does support
8764 floating-point instructions. The @option{-msoft-float} option indicates
8765 that floating-point support should not be assumed.
8767 @item -mleaf-procedures
8768 @itemx -mno-leaf-procedures
8769 @opindex mleaf-procedures
8770 @opindex mno-leaf-procedures
8771 Do (or do not) attempt to alter leaf procedures to be callable with the
8772 @code{bal} instruction as well as @code{call}. This will result in more
8773 efficient code for explicit calls when the @code{bal} instruction can be
8774 substituted by the assembler or linker, but less efficient code in other
8775 cases, such as calls via function pointers, or using a linker that doesn't
8776 support this optimization.
8779 @itemx -mno-tail-call
8781 @opindex mno-tail-call
8782 Do (or do not) make additional attempts (beyond those of the
8783 machine-independent portions of the compiler) to optimize tail-recursive
8784 calls into branches. You may not want to do this because the detection of
8785 cases where this is not valid is not totally complete. The default is
8786 @option{-mno-tail-call}.
8788 @item -mcomplex-addr
8789 @itemx -mno-complex-addr
8790 @opindex mcomplex-addr
8791 @opindex mno-complex-addr
8792 Assume (or do not assume) that the use of a complex addressing mode is a
8793 win on this implementation of the i960. Complex addressing modes may not
8794 be worthwhile on the K-series, but they definitely are on the C-series.
8795 The default is currently @option{-mcomplex-addr} for all processors except
8799 @itemx -mno-code-align
8800 @opindex mcode-align
8801 @opindex mno-code-align
8802 Align code to 8-byte boundaries for faster fetching (or don't bother).
8803 Currently turned on by default for C-series implementations only.
8806 @item -mclean-linkage
8807 @itemx -mno-clean-linkage
8808 @opindex mclean-linkage
8809 @opindex mno-clean-linkage
8810 These options are not fully implemented.
8814 @itemx -mic2.0-compat
8815 @itemx -mic3.0-compat
8817 @opindex mic2.0-compat
8818 @opindex mic3.0-compat
8819 Enable compatibility with iC960 v2.0 or v3.0.
8823 @opindex masm-compat
8825 Enable compatibility with the iC960 assembler.
8827 @item -mstrict-align
8828 @itemx -mno-strict-align
8829 @opindex mstrict-align
8830 @opindex mno-strict-align
8831 Do not permit (do permit) unaligned accesses.
8835 Enable structure-alignment compatibility with Intel's gcc release version
8836 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8838 @item -mlong-double-64
8839 @opindex mlong-double-64
8840 Implement type @samp{long double} as 64-bit floating point numbers.
8841 Without the option @samp{long double} is implemented by 80-bit
8842 floating point numbers. The only reason we have it because there is
8843 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8844 is only useful for people using soft-float targets. Otherwise, we
8845 should recommend against use of it.
8849 @node DEC Alpha Options
8850 @subsection DEC Alpha Options
8852 These @samp{-m} options are defined for the DEC Alpha implementations:
8855 @item -mno-soft-float
8857 @opindex mno-soft-float
8858 @opindex msoft-float
8859 Use (do not use) the hardware floating-point instructions for
8860 floating-point operations. When @option{-msoft-float} is specified,
8861 functions in @file{libgcc.a} will be used to perform floating-point
8862 operations. Unless they are replaced by routines that emulate the
8863 floating-point operations, or compiled in such a way as to call such
8864 emulations routines, these routines will issue floating-point
8865 operations. If you are compiling for an Alpha without floating-point
8866 operations, you must ensure that the library is built so as not to call
8869 Note that Alpha implementations without floating-point operations are
8870 required to have floating-point registers.
8875 @opindex mno-fp-regs
8876 Generate code that uses (does not use) the floating-point register set.
8877 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8878 register set is not used, floating point operands are passed in integer
8879 registers as if they were integers and floating-point results are passed
8880 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8881 so any function with a floating-point argument or return value called by code
8882 compiled with @option{-mno-fp-regs} must also be compiled with that
8885 A typical use of this option is building a kernel that does not use,
8886 and hence need not save and restore, any floating-point registers.
8890 The Alpha architecture implements floating-point hardware optimized for
8891 maximum performance. It is mostly compliant with the IEEE floating
8892 point standard. However, for full compliance, software assistance is
8893 required. This option generates code fully IEEE compliant code
8894 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8895 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8896 defined during compilation. The resulting code is less efficient but is
8897 able to correctly support denormalized numbers and exceptional IEEE
8898 values such as not-a-number and plus/minus infinity. Other Alpha
8899 compilers call this option @option{-ieee_with_no_inexact}.
8901 @item -mieee-with-inexact
8902 @opindex mieee-with-inexact
8903 This is like @option{-mieee} except the generated code also maintains
8904 the IEEE @var{inexact-flag}. Turning on this option causes the
8905 generated code to implement fully-compliant IEEE math. In addition to
8906 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8907 macro. On some Alpha implementations the resulting code may execute
8908 significantly slower than the code generated by default. Since there is
8909 very little code that depends on the @var{inexact-flag}, you should
8910 normally not specify this option. Other Alpha compilers call this
8911 option @option{-ieee_with_inexact}.
8913 @item -mfp-trap-mode=@var{trap-mode}
8914 @opindex mfp-trap-mode
8915 This option controls what floating-point related traps are enabled.
8916 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8917 The trap mode can be set to one of four values:
8921 This is the default (normal) setting. The only traps that are enabled
8922 are the ones that cannot be disabled in software (e.g., division by zero
8926 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8930 Like @samp{su}, but the instructions are marked to be safe for software
8931 completion (see Alpha architecture manual for details).
8934 Like @samp{su}, but inexact traps are enabled as well.
8937 @item -mfp-rounding-mode=@var{rounding-mode}
8938 @opindex mfp-rounding-mode
8939 Selects the IEEE rounding mode. Other Alpha compilers call this option
8940 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8945 Normal IEEE rounding mode. Floating point numbers are rounded towards
8946 the nearest machine number or towards the even machine number in case
8950 Round towards minus infinity.
8953 Chopped rounding mode. Floating point numbers are rounded towards zero.
8956 Dynamic rounding mode. A field in the floating point control register
8957 (@var{fpcr}, see Alpha architecture reference manual) controls the
8958 rounding mode in effect. The C library initializes this register for
8959 rounding towards plus infinity. Thus, unless your program modifies the
8960 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8963 @item -mtrap-precision=@var{trap-precision}
8964 @opindex mtrap-precision
8965 In the Alpha architecture, floating point traps are imprecise. This
8966 means without software assistance it is impossible to recover from a
8967 floating trap and program execution normally needs to be terminated.
8968 GCC can generate code that can assist operating system trap handlers
8969 in determining the exact location that caused a floating point trap.
8970 Depending on the requirements of an application, different levels of
8971 precisions can be selected:
8975 Program precision. This option is the default and means a trap handler
8976 can only identify which program caused a floating point exception.
8979 Function precision. The trap handler can determine the function that
8980 caused a floating point exception.
8983 Instruction precision. The trap handler can determine the exact
8984 instruction that caused a floating point exception.
8987 Other Alpha compilers provide the equivalent options called
8988 @option{-scope_safe} and @option{-resumption_safe}.
8990 @item -mieee-conformant
8991 @opindex mieee-conformant
8992 This option marks the generated code as IEEE conformant. You must not
8993 use this option unless you also specify @option{-mtrap-precision=i} and either
8994 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8995 is to emit the line @samp{.eflag 48} in the function prologue of the
8996 generated assembly file. Under DEC Unix, this has the effect that
8997 IEEE-conformant math library routines will be linked in.
8999 @item -mbuild-constants
9000 @opindex mbuild-constants
9001 Normally GCC examines a 32- or 64-bit integer constant to
9002 see if it can construct it from smaller constants in two or three
9003 instructions. If it cannot, it will output the constant as a literal and
9004 generate code to load it from the data segment at runtime.
9006 Use this option to require GCC to construct @emph{all} integer constants
9007 using code, even if it takes more instructions (the maximum is six).
9009 You would typically use this option to build a shared library dynamic
9010 loader. Itself a shared library, it must relocate itself in memory
9011 before it can find the variables and constants in its own data segment.
9017 Select whether to generate code to be assembled by the vendor-supplied
9018 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9036 Indicate whether GCC should generate code to use the optional BWX,
9037 CIX, FIX and MAX instruction sets. The default is to use the instruction
9038 sets supported by the CPU type specified via @option{-mcpu=} option or that
9039 of the CPU on which GCC was built if none was specified.
9044 @opindex mfloat-ieee
9045 Generate code that uses (does not use) VAX F and G floating point
9046 arithmetic instead of IEEE single and double precision.
9048 @item -mexplicit-relocs
9049 @itemx -mno-explicit-relocs
9050 @opindex mexplicit-relocs
9051 @opindex mno-explicit-relocs
9052 Older Alpha assemblers provided no way to generate symbol relocations
9053 except via assembler macros. Use of these macros does not allow
9054 optimal instruction scheduling. GNU binutils as of version 2.12
9055 supports a new syntax that allows the compiler to explicitly mark
9056 which relocations should apply to which instructions. This option
9057 is mostly useful for debugging, as GCC detects the capabilities of
9058 the assembler when it is built and sets the default accordingly.
9062 @opindex msmall-data
9063 @opindex mlarge-data
9064 When @option{-mexplicit-relocs} is in effect, static data is
9065 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9066 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9067 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9068 16-bit relocations off of the @code{$gp} register. This limits the
9069 size of the small data area to 64KB, but allows the variables to be
9070 directly accessed via a single instruction.
9072 The default is @option{-mlarge-data}. With this option the data area
9073 is limited to just below 2GB. Programs that require more than 2GB of
9074 data must use @code{malloc} or @code{mmap} to allocate the data in the
9075 heap instead of in the program's data segment.
9077 When generating code for shared libraries, @option{-fpic} implies
9078 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9082 @opindex msmall-text
9083 @opindex mlarge-text
9084 When @option{-msmall-text} is used, the compiler assumes that the
9085 code of the entire program (or shared library) fits in 4MB, and is
9086 thus reachable with a branch instruction. When @option{-msmall-data}
9087 is used, the compiler can assume that all local symbols share the
9088 same @code{$gp} value, and thus reduce the number of instructions
9089 required for a function call from 4 to 1.
9091 The default is @option{-mlarge-text}.
9093 @item -mcpu=@var{cpu_type}
9095 Set the instruction set and instruction scheduling parameters for
9096 machine type @var{cpu_type}. You can specify either the @samp{EV}
9097 style name or the corresponding chip number. GCC supports scheduling
9098 parameters for the EV4, EV5 and EV6 family of processors and will
9099 choose the default values for the instruction set from the processor
9100 you specify. If you do not specify a processor type, GCC will default
9101 to the processor on which the compiler was built.
9103 Supported values for @var{cpu_type} are
9109 Schedules as an EV4 and has no instruction set extensions.
9113 Schedules as an EV5 and has no instruction set extensions.
9117 Schedules as an EV5 and supports the BWX extension.
9122 Schedules as an EV5 and supports the BWX and MAX extensions.
9126 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9130 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9133 @item -mtune=@var{cpu_type}
9135 Set only the instruction scheduling parameters for machine type
9136 @var{cpu_type}. The instruction set is not changed.
9138 @item -mmemory-latency=@var{time}
9139 @opindex mmemory-latency
9140 Sets the latency the scheduler should assume for typical memory
9141 references as seen by the application. This number is highly
9142 dependent on the memory access patterns used by the application
9143 and the size of the external cache on the machine.
9145 Valid options for @var{time} are
9149 A decimal number representing clock cycles.
9155 The compiler contains estimates of the number of clock cycles for
9156 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9157 (also called Dcache, Scache, and Bcache), as well as to main memory.
9158 Note that L3 is only valid for EV5.
9163 @node DEC Alpha/VMS Options
9164 @subsection DEC Alpha/VMS Options
9166 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9169 @item -mvms-return-codes
9170 @opindex mvms-return-codes
9171 Return VMS condition codes from main. The default is to return POSIX
9172 style condition (e.g.@ error) codes.
9175 @node H8/300 Options
9176 @subsection H8/300 Options
9178 These @samp{-m} options are defined for the H8/300 implementations:
9183 Shorten some address references at link time, when possible; uses the
9184 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9185 ld.info, Using ld}, for a fuller description.
9189 Generate code for the H8/300H@.
9193 Generate code for the H8S@.
9197 Generate code for the H8S and H8/300H in the normal mode. This switch
9198 must be used either with -mh or -ms.
9202 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9206 Make @code{int} data 32 bits by default.
9210 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9211 The default for the H8/300H and H8S is to align longs and floats on 4
9213 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9214 This option has no effect on the H8/300.
9218 @subsection SH Options
9220 These @samp{-m} options are defined for the SH implementations:
9225 Generate code for the SH1.
9229 Generate code for the SH2.
9232 Generate code for the SH2e.
9236 Generate code for the SH3.
9240 Generate code for the SH3e.
9244 Generate code for the SH4 without a floating-point unit.
9246 @item -m4-single-only
9247 @opindex m4-single-only
9248 Generate code for the SH4 with a floating-point unit that only
9249 supports single-precision arithmetic.
9253 Generate code for the SH4 assuming the floating-point unit is in
9254 single-precision mode by default.
9258 Generate code for the SH4.
9262 Compile code for the processor in big endian mode.
9266 Compile code for the processor in little endian mode.
9270 Align doubles at 64-bit boundaries. Note that this changes the calling
9271 conventions, and thus some functions from the standard C library will
9272 not work unless you recompile it first with @option{-mdalign}.
9276 Shorten some address references at link time, when possible; uses the
9277 linker option @option{-relax}.
9281 Use 32-bit offsets in @code{switch} tables. The default is to use
9286 Enable the use of the instruction @code{fmovd}.
9290 Comply with the calling conventions defined by Renesas.
9294 Mark the @code{MAC} register as call-clobbered, even if
9295 @option{-mhitachi} is given.
9299 Increase IEEE-compliance of floating-point code.
9303 Dump instruction size and location in the assembly code.
9307 This option is deprecated. It pads structures to multiple of 4 bytes,
9308 which is incompatible with the SH ABI@.
9312 Optimize for space instead of speed. Implied by @option{-Os}.
9316 When generating position-independent code, emit function calls using
9317 the Global Offset Table instead of the Procedure Linkage Table.
9321 Generate a library function call to invalidate instruction cache
9322 entries, after fixing up a trampoline. This library function call
9323 doesn't assume it can write to the whole memory address space. This
9324 is the default when the target is @code{sh-*-linux*}.
9327 @node System V Options
9328 @subsection Options for System V
9330 These additional options are available on System V Release 4 for
9331 compatibility with other compilers on those systems:
9336 Create a shared object.
9337 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9341 Identify the versions of each tool used by the compiler, in a
9342 @code{.ident} assembler directive in the output.
9346 Refrain from adding @code{.ident} directives to the output file (this is
9349 @item -YP,@var{dirs}
9351 Search the directories @var{dirs}, and no others, for libraries
9352 specified with @option{-l}.
9356 Look in the directory @var{dir} to find the M4 preprocessor.
9357 The assembler uses this option.
9358 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9359 @c the generic assembler that comes with Solaris takes just -Ym.
9362 @node TMS320C3x/C4x Options
9363 @subsection TMS320C3x/C4x Options
9364 @cindex TMS320C3x/C4x Options
9366 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9370 @item -mcpu=@var{cpu_type}
9372 Set the instruction set, register set, and instruction scheduling
9373 parameters for machine type @var{cpu_type}. Supported values for
9374 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9375 @samp{c44}. The default is @samp{c40} to generate code for the
9380 @itemx -msmall-memory
9382 @opindex mbig-memory
9384 @opindex msmall-memory
9386 Generates code for the big or small memory model. The small memory
9387 model assumed that all data fits into one 64K word page. At run-time
9388 the data page (DP) register must be set to point to the 64K page
9389 containing the .bss and .data program sections. The big memory model is
9390 the default and requires reloading of the DP register for every direct
9397 Allow (disallow) allocation of general integer operands into the block
9404 Enable (disable) generation of code using decrement and branch,
9405 DBcond(D), instructions. This is enabled by default for the C4x. To be
9406 on the safe side, this is disabled for the C3x, since the maximum
9407 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9408 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9409 that it can utilize the decrement and branch instruction, but will give
9410 up if there is more than one memory reference in the loop. Thus a loop
9411 where the loop counter is decremented can generate slightly more
9412 efficient code, in cases where the RPTB instruction cannot be utilized.
9414 @item -mdp-isr-reload
9416 @opindex mdp-isr-reload
9418 Force the DP register to be saved on entry to an interrupt service
9419 routine (ISR), reloaded to point to the data section, and restored on
9420 exit from the ISR@. This should not be required unless someone has
9421 violated the small memory model by modifying the DP register, say within
9428 For the C3x use the 24-bit MPYI instruction for integer multiplies
9429 instead of a library call to guarantee 32-bit results. Note that if one
9430 of the operands is a constant, then the multiplication will be performed
9431 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9432 then squaring operations are performed inline instead of a library call.
9435 @itemx -mno-fast-fix
9437 @opindex mno-fast-fix
9438 The C3x/C4x FIX instruction to convert a floating point value to an
9439 integer value chooses the nearest integer less than or equal to the
9440 floating point value rather than to the nearest integer. Thus if the
9441 floating point number is negative, the result will be incorrectly
9442 truncated an additional code is necessary to detect and correct this
9443 case. This option can be used to disable generation of the additional
9444 code required to correct the result.
9450 Enable (disable) generation of repeat block sequences using the RPTB
9451 instruction for zero overhead looping. The RPTB construct is only used
9452 for innermost loops that do not call functions or jump across the loop
9453 boundaries. There is no advantage having nested RPTB loops due to the
9454 overhead required to save and restore the RC, RS, and RE registers.
9455 This is enabled by default with @option{-O2}.
9457 @item -mrpts=@var{count}
9461 Enable (disable) the use of the single instruction repeat instruction
9462 RPTS@. If a repeat block contains a single instruction, and the loop
9463 count can be guaranteed to be less than the value @var{count}, GCC will
9464 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9465 then a RPTS will be emitted even if the loop count cannot be determined
9466 at compile time. Note that the repeated instruction following RPTS does
9467 not have to be reloaded from memory each iteration, thus freeing up the
9468 CPU buses for operands. However, since interrupts are blocked by this
9469 instruction, it is disabled by default.
9471 @item -mloop-unsigned
9472 @itemx -mno-loop-unsigned
9473 @opindex mloop-unsigned
9474 @opindex mno-loop-unsigned
9475 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9476 is @math{2^{31} + 1} since these instructions test if the iteration count is
9477 negative to terminate the loop. If the iteration count is unsigned
9478 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9479 exceeded. This switch allows an unsigned iteration count.
9483 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9484 with. This also enforces compatibility with the API employed by the TI
9485 C3x C compiler. For example, long doubles are passed as structures
9486 rather than in floating point registers.
9492 Generate code that uses registers (stack) for passing arguments to functions.
9493 By default, arguments are passed in registers where possible rather
9494 than by pushing arguments on to the stack.
9496 @item -mparallel-insns
9497 @itemx -mno-parallel-insns
9498 @opindex mparallel-insns
9499 @opindex mno-parallel-insns
9500 Allow the generation of parallel instructions. This is enabled by
9501 default with @option{-O2}.
9503 @item -mparallel-mpy
9504 @itemx -mno-parallel-mpy
9505 @opindex mparallel-mpy
9506 @opindex mno-parallel-mpy
9507 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9508 provided @option{-mparallel-insns} is also specified. These instructions have
9509 tight register constraints which can pessimize the code generation
9515 @subsection V850 Options
9516 @cindex V850 Options
9518 These @samp{-m} options are defined for V850 implementations:
9522 @itemx -mno-long-calls
9523 @opindex mlong-calls
9524 @opindex mno-long-calls
9525 Treat all calls as being far away (near). If calls are assumed to be
9526 far away, the compiler will always load the functions address up into a
9527 register, and call indirect through the pointer.
9533 Do not optimize (do optimize) basic blocks that use the same index
9534 pointer 4 or more times to copy pointer into the @code{ep} register, and
9535 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9536 option is on by default if you optimize.
9538 @item -mno-prolog-function
9539 @itemx -mprolog-function
9540 @opindex mno-prolog-function
9541 @opindex mprolog-function
9542 Do not use (do use) external functions to save and restore registers at
9543 the prolog and epilog of a function. The external functions are slower,
9544 but use less code space if more than one function saves the same number
9545 of registers. The @option{-mprolog-function} option is on by default if
9550 Try to make the code as small as possible. At present, this just turns
9551 on the @option{-mep} and @option{-mprolog-function} options.
9555 Put static or global variables whose size is @var{n} bytes or less into
9556 the tiny data area that register @code{ep} points to. The tiny data
9557 area can hold up to 256 bytes in total (128 bytes for byte references).
9561 Put static or global variables whose size is @var{n} bytes or less into
9562 the small data area that register @code{gp} points to. The small data
9563 area can hold up to 64 kilobytes.
9567 Put static or global variables whose size is @var{n} bytes or less into
9568 the first 32 kilobytes of memory.
9572 Specify that the target processor is the V850.
9575 @opindex mbig-switch
9576 Generate code suitable for big switch tables. Use this option only if
9577 the assembler/linker complain about out of range branches within a switch
9582 This option will cause r2 and r5 to be used in the code generated by
9583 the compiler. This setting is the default.
9586 @opindex mno-app-regs
9587 This option will cause r2 and r5 to be treated as fixed registers.
9591 Specify that the target processor is the V850E. The preprocessor
9592 constant @samp{__v850e__} will be defined if this option is used.
9594 If neither @option{-mv850} nor @option{-mv850e} are defined
9595 then a default target processor will be chosen and the relevant
9596 @samp{__v850*__} preprocessor constant will be defined.
9598 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9599 defined, regardless of which processor variant is the target.
9601 @item -mdisable-callt
9602 @opindex mdisable-callt
9603 This option will suppress generation of the CALLT instruction for the
9604 v850e flavors of the v850 architecture. The default is
9605 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9610 @subsection ARC Options
9613 These options are defined for ARC implementations:
9618 Compile code for little endian mode. This is the default.
9622 Compile code for big endian mode.
9625 @opindex mmangle-cpu
9626 Prepend the name of the cpu to all public symbol names.
9627 In multiple-processor systems, there are many ARC variants with different
9628 instruction and register set characteristics. This flag prevents code
9629 compiled for one cpu to be linked with code compiled for another.
9630 No facility exists for handling variants that are ``almost identical''.
9631 This is an all or nothing option.
9633 @item -mcpu=@var{cpu}
9635 Compile code for ARC variant @var{cpu}.
9636 Which variants are supported depend on the configuration.
9637 All variants support @option{-mcpu=base}, this is the default.
9639 @item -mtext=@var{text-section}
9640 @itemx -mdata=@var{data-section}
9641 @itemx -mrodata=@var{readonly-data-section}
9645 Put functions, data, and readonly data in @var{text-section},
9646 @var{data-section}, and @var{readonly-data-section} respectively
9647 by default. This can be overridden with the @code{section} attribute.
9648 @xref{Variable Attributes}.
9653 @subsection NS32K Options
9654 @cindex NS32K options
9656 These are the @samp{-m} options defined for the 32000 series. The default
9657 values for these options depends on which style of 32000 was selected when
9658 the compiler was configured; the defaults for the most common choices are
9666 Generate output for a 32032. This is the default
9667 when the compiler is configured for 32032 and 32016 based systems.
9673 Generate output for a 32332. This is the default
9674 when the compiler is configured for 32332-based systems.
9680 Generate output for a 32532. This is the default
9681 when the compiler is configured for 32532-based systems.
9685 Generate output containing 32081 instructions for floating point.
9686 This is the default for all systems.
9690 Generate output containing 32381 instructions for floating point. This
9691 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9692 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9696 Try and generate multiply-add floating point instructions @code{polyF}
9697 and @code{dotF}. This option is only available if the @option{-m32381}
9698 option is in effect. Using these instructions requires changes to
9699 register allocation which generally has a negative impact on
9700 performance. This option should only be enabled when compiling code
9701 particularly likely to make heavy use of multiply-add instructions.
9704 @opindex mnomulti-add
9705 Do not try and generate multiply-add floating point instructions
9706 @code{polyF} and @code{dotF}. This is the default on all platforms.
9709 @opindex msoft-float
9710 Generate output containing library calls for floating point.
9711 @strong{Warning:} the requisite libraries may not be available.
9713 @item -mieee-compare
9714 @itemx -mno-ieee-compare
9715 @opindex mieee-compare
9716 @opindex mno-ieee-compare
9717 Control whether or not the compiler uses IEEE floating point
9718 comparisons. These handle correctly the case where the result of a
9719 comparison is unordered.
9720 @strong{Warning:} the requisite kernel support may not be available.
9723 @opindex mnobitfield
9724 Do not use the bit-field instructions. On some machines it is faster to
9725 use shifting and masking operations. This is the default for the pc532.
9729 Do use the bit-field instructions. This is the default for all platforms
9734 Use a different function-calling convention, in which functions
9735 that take a fixed number of arguments return pop their
9736 arguments on return with the @code{ret} instruction.
9738 This calling convention is incompatible with the one normally
9739 used on Unix, so you cannot use it if you need to call libraries
9740 compiled with the Unix compiler.
9742 Also, you must provide function prototypes for all functions that
9743 take variable numbers of arguments (including @code{printf});
9744 otherwise incorrect code will be generated for calls to those
9747 In addition, seriously incorrect code will result if you call a
9748 function with too many arguments. (Normally, extra arguments are
9749 harmlessly ignored.)
9751 This option takes its name from the 680x0 @code{rtd} instruction.
9756 Use a different function-calling convention where the first two arguments
9757 are passed in registers.
9759 This calling convention is incompatible with the one normally
9760 used on Unix, so you cannot use it if you need to call libraries
9761 compiled with the Unix compiler.
9764 @opindex mnoregparam
9765 Do not pass any arguments in registers. This is the default for all
9770 It is OK to use the sb as an index register which is always loaded with
9771 zero. This is the default for the pc532-netbsd target.
9775 The sb register is not available for use or has not been initialized to
9776 zero by the run time system. This is the default for all targets except
9777 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9778 @option{-fpic} is set.
9782 Many ns32000 series addressing modes use displacements of up to 512MB@.
9783 If an address is above 512MB then displacements from zero can not be used.
9784 This option causes code to be generated which can be loaded above 512MB@.
9785 This may be useful for operating systems or ROM code.
9789 Assume code will be loaded in the first 512MB of virtual address space.
9790 This is the default for all platforms.
9796 @subsection AVR Options
9799 These options are defined for AVR implementations:
9802 @item -mmcu=@var{mcu}
9804 Specify ATMEL AVR instruction set or MCU type.
9806 Instruction set avr1 is for the minimal AVR core, not supported by the C
9807 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9808 attiny11, attiny12, attiny15, attiny28).
9810 Instruction set avr2 (default) is for the classic AVR core with up to
9811 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9812 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9813 at90c8534, at90s8535).
9815 Instruction set avr3 is for the classic AVR core with up to 128K program
9816 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9818 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9819 memory space (MCU types: atmega8, atmega83, atmega85).
9821 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9822 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9823 atmega64, atmega128, at43usb355, at94k).
9827 Output instruction sizes to the asm file.
9829 @item -minit-stack=@var{N}
9830 @opindex minit-stack
9831 Specify the initial stack address, which may be a symbol or numeric value,
9832 @samp{__stack} is the default.
9834 @item -mno-interrupts
9835 @opindex mno-interrupts
9836 Generated code is not compatible with hardware interrupts.
9837 Code size will be smaller.
9839 @item -mcall-prologues
9840 @opindex mcall-prologues
9841 Functions prologues/epilogues expanded as call to appropriate
9842 subroutines. Code size will be smaller.
9844 @item -mno-tablejump
9845 @opindex mno-tablejump
9846 Do not generate tablejump insns which sometimes increase code size.
9849 @opindex mtiny-stack
9850 Change only the low 8 bits of the stack pointer.
9854 @subsection MCore Options
9855 @cindex MCore options
9857 These are the @samp{-m} options defined for the Motorola M*Core
9865 @opindex mno-hardlit
9866 Inline constants into the code stream if it can be done in two
9867 instructions or less.
9873 Use the divide instruction. (Enabled by default).
9875 @item -mrelax-immediate
9876 @itemx -mno-relax-immediate
9877 @opindex mrelax-immediate
9878 @opindex mno-relax-immediate
9879 Allow arbitrary sized immediates in bit operations.
9881 @item -mwide-bitfields
9882 @itemx -mno-wide-bitfields
9883 @opindex mwide-bitfields
9884 @opindex mno-wide-bitfields
9885 Always treat bit-fields as int-sized.
9887 @item -m4byte-functions
9888 @itemx -mno-4byte-functions
9889 @opindex m4byte-functions
9890 @opindex mno-4byte-functions
9891 Force all functions to be aligned to a four byte boundary.
9893 @item -mcallgraph-data
9894 @itemx -mno-callgraph-data
9895 @opindex mcallgraph-data
9896 @opindex mno-callgraph-data
9897 Emit callgraph information.
9900 @itemx -mno-slow-bytes
9901 @opindex mslow-bytes
9902 @opindex mno-slow-bytes
9903 Prefer word access when reading byte quantities.
9905 @item -mlittle-endian
9907 @opindex mlittle-endian
9908 @opindex mbig-endian
9909 Generate code for a little endian target.
9915 Generate code for the 210 processor.
9919 @subsection IA-64 Options
9920 @cindex IA-64 Options
9922 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9926 @opindex mbig-endian
9927 Generate code for a big endian target. This is the default for HP-UX@.
9929 @item -mlittle-endian
9930 @opindex mlittle-endian
9931 Generate code for a little endian target. This is the default for AIX5
9938 Generate (or don't) code for the GNU assembler. This is the default.
9939 @c Also, this is the default if the configure option @option{--with-gnu-as}
9946 Generate (or don't) code for the GNU linker. This is the default.
9947 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9952 Generate code that does not use a global pointer register. The result
9953 is not position independent code, and violates the IA-64 ABI@.
9955 @item -mvolatile-asm-stop
9956 @itemx -mno-volatile-asm-stop
9957 @opindex mvolatile-asm-stop
9958 @opindex mno-volatile-asm-stop
9959 Generate (or don't) a stop bit immediately before and after volatile asm
9964 Generate code that works around Itanium B step errata.
9966 @item -mregister-names
9967 @itemx -mno-register-names
9968 @opindex mregister-names
9969 @opindex mno-register-names
9970 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9971 the stacked registers. This may make assembler output more readable.
9977 Disable (or enable) optimizations that use the small data section. This may
9978 be useful for working around optimizer bugs.
9981 @opindex mconstant-gp
9982 Generate code that uses a single constant global pointer value. This is
9983 useful when compiling kernel code.
9987 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9988 This is useful when compiling firmware code.
9990 @item -minline-float-divide-min-latency
9991 @opindex minline-float-divide-min-latency
9992 Generate code for inline divides of floating point values
9993 using the minimum latency algorithm.
9995 @item -minline-float-divide-max-throughput
9996 @opindex minline-float-divide-max-throughput
9997 Generate code for inline divides of floating point values
9998 using the maximum throughput algorithm.
10000 @item -minline-int-divide-min-latency
10001 @opindex minline-int-divide-min-latency
10002 Generate code for inline divides of integer values
10003 using the minimum latency algorithm.
10005 @item -minline-int-divide-max-throughput
10006 @opindex minline-int-divide-max-throughput
10007 Generate code for inline divides of integer values
10008 using the maximum throughput algorithm.
10010 @item -mno-dwarf2-asm
10011 @itemx -mdwarf2-asm
10012 @opindex mno-dwarf2-asm
10013 @opindex mdwarf2-asm
10014 Don't (or do) generate assembler code for the DWARF2 line number debugging
10015 info. This may be useful when not using the GNU assembler.
10017 @item -mfixed-range=@var{register-range}
10018 @opindex mfixed-range
10019 Generate code treating the given register range as fixed registers.
10020 A fixed register is one that the register allocator can not use. This is
10021 useful when compiling kernel code. A register range is specified as
10022 two registers separated by a dash. Multiple register ranges can be
10023 specified separated by a comma.
10025 @item -mearly-stop-bits
10026 @itemx -mno-early-stop-bits
10027 @opindex mearly-stop-bits
10028 @opindex mno-early-stop-bits
10029 Allow stop bits to be placed earlier than immediately preceding the
10030 instruction that triggered the stop bit. This can improve instruction
10031 scheduling, but does not always do so.
10035 @subsection D30V Options
10036 @cindex D30V Options
10038 These @samp{-m} options are defined for D30V implementations:
10043 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10044 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10045 memory, which starts at location @code{0x80000000}.
10048 @opindex mextmemory
10049 Same as the @option{-mextmem} switch.
10053 Link the @samp{.text} section into onchip text memory, which starts at
10054 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10055 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10056 into onchip data memory, which starts at location @code{0x20000000}.
10058 @item -mno-asm-optimize
10059 @itemx -masm-optimize
10060 @opindex mno-asm-optimize
10061 @opindex masm-optimize
10062 Disable (enable) passing @option{-O} to the assembler when optimizing.
10063 The assembler uses the @option{-O} option to automatically parallelize
10064 adjacent short instructions where possible.
10066 @item -mbranch-cost=@var{n}
10067 @opindex mbranch-cost
10068 Increase the internal costs of branches to @var{n}. Higher costs means
10069 that the compiler will issue more instructions to avoid doing a branch.
10072 @item -mcond-exec=@var{n}
10073 @opindex mcond-exec
10074 Specify the maximum number of conditionally executed instructions that
10075 replace a branch. The default is 4.
10078 @node S/390 and zSeries Options
10079 @subsection S/390 and zSeries Options
10080 @cindex S/390 and zSeries Options
10082 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10086 @itemx -msoft-float
10087 @opindex mhard-float
10088 @opindex msoft-float
10089 Use (do not use) the hardware floating-point instructions and registers
10090 for floating-point operations. When @option{-msoft-float} is specified,
10091 functions in @file{libgcc.a} will be used to perform floating-point
10092 operations. When @option{-mhard-float} is specified, the compiler
10093 generates IEEE floating-point instructions. This is the default.
10096 @itemx -mno-backchain
10097 @opindex mbackchain
10098 @opindex mno-backchain
10099 Generate (or do not generate) code which maintains an explicit
10100 backchain within the stack frame that points to the caller's frame.
10101 This is currently needed to allow debugging. The default is to
10102 generate the backchain.
10105 @itemx -mno-small-exec
10106 @opindex msmall-exec
10107 @opindex mno-small-exec
10108 Generate (or do not generate) code using the @code{bras} instruction
10109 to do subroutine calls.
10110 This only works reliably if the total executable size does not
10111 exceed 64k. The default is to use the @code{basr} instruction instead,
10112 which does not have this limitation.
10118 When @option{-m31} is specified, generate code compliant to the
10119 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10120 code compliant to the Linux for zSeries ABI@. This allows GCC in
10121 particular to generate 64-bit instructions. For the @samp{s390}
10122 targets, the default is @option{-m31}, while the @samp{s390x}
10123 targets default to @option{-m64}.
10129 When @option{-mzarch} is specified, generate code using the
10130 instructions available on z/Architecture.
10131 When @option{-mesa} is specified, generate code using the
10132 instructions available on ESA/390. Note that @option{-mesa} is
10133 not possible with @option{-m64}.
10134 When generating code compliant to the Linux for S/390 ABI,
10135 the default is @option{-mesa}. When generating code compliant
10136 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10142 Generate (or do not generate) code using the @code{mvcle} instruction
10143 to perform block moves. When @option{-mno-mvcle} is specified,
10144 use a @code{mvc} loop instead. This is the default.
10150 Print (or do not print) additional debug information when compiling.
10151 The default is to not print debug information.
10153 @item -march=@var{cpu-type}
10155 Generate code that will run on @var{cpu-type}, which is the name of a system
10156 representing a certain processor type. Possible values for
10157 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10158 When generating code using the instructions available on z/Architecture,
10159 the default is @option{-march=z900}. Otherwise, the default is
10160 @option{-march=g5}.
10162 @item -mtune=@var{cpu-type}
10164 Tune to @var{cpu-type} everything applicable about the generated code,
10165 except for the ABI and the set of available instructions.
10166 The list of @var{cpu-type} values is the same as for @option{-march}.
10167 The default is the value used for @option{-march}.
10172 @subsection CRIS Options
10173 @cindex CRIS Options
10175 These options are defined specifically for the CRIS ports.
10178 @item -march=@var{architecture-type}
10179 @itemx -mcpu=@var{architecture-type}
10182 Generate code for the specified architecture. The choices for
10183 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10184 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10185 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10188 @item -mtune=@var{architecture-type}
10190 Tune to @var{architecture-type} everything applicable about the generated
10191 code, except for the ABI and the set of available instructions. The
10192 choices for @var{architecture-type} are the same as for
10193 @option{-march=@var{architecture-type}}.
10195 @item -mmax-stack-frame=@var{n}
10196 @opindex mmax-stack-frame
10197 Warn when the stack frame of a function exceeds @var{n} bytes.
10199 @item -melinux-stacksize=@var{n}
10200 @opindex melinux-stacksize
10201 Only available with the @samp{cris-axis-aout} target. Arranges for
10202 indications in the program to the kernel loader that the stack of the
10203 program should be set to @var{n} bytes.
10209 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10210 @option{-march=v3} and @option{-march=v8} respectively.
10214 Enable CRIS-specific verbose debug-related information in the assembly
10215 code. This option also has the effect to turn off the @samp{#NO_APP}
10216 formatted-code indicator to the assembler at the beginning of the
10221 Do not use condition-code results from previous instruction; always emit
10222 compare and test instructions before use of condition codes.
10224 @item -mno-side-effects
10225 @opindex mno-side-effects
10226 Do not emit instructions with side-effects in addressing modes other than
10229 @item -mstack-align
10230 @itemx -mno-stack-align
10231 @itemx -mdata-align
10232 @itemx -mno-data-align
10233 @itemx -mconst-align
10234 @itemx -mno-const-align
10235 @opindex mstack-align
10236 @opindex mno-stack-align
10237 @opindex mdata-align
10238 @opindex mno-data-align
10239 @opindex mconst-align
10240 @opindex mno-const-align
10241 These options (no-options) arranges (eliminate arrangements) for the
10242 stack-frame, individual data and constants to be aligned for the maximum
10243 single data access size for the chosen CPU model. The default is to
10244 arrange for 32-bit alignment. ABI details such as structure layout are
10245 not affected by these options.
10253 Similar to the stack- data- and const-align options above, these options
10254 arrange for stack-frame, writable data and constants to all be 32-bit,
10255 16-bit or 8-bit aligned. The default is 32-bit alignment.
10257 @item -mno-prologue-epilogue
10258 @itemx -mprologue-epilogue
10259 @opindex mno-prologue-epilogue
10260 @opindex mprologue-epilogue
10261 With @option{-mno-prologue-epilogue}, the normal function prologue and
10262 epilogue that sets up the stack-frame are omitted and no return
10263 instructions or return sequences are generated in the code. Use this
10264 option only together with visual inspection of the compiled code: no
10265 warnings or errors are generated when call-saved registers must be saved,
10266 or storage for local variable needs to be allocated.
10270 @opindex mno-gotplt
10272 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10273 instruction sequences that load addresses for functions from the PLT part
10274 of the GOT rather than (traditional on other architectures) calls to the
10275 PLT. The default is @option{-mgotplt}.
10279 Legacy no-op option only recognized with the cris-axis-aout target.
10283 Legacy no-op option only recognized with the cris-axis-elf and
10284 cris-axis-linux-gnu targets.
10288 Only recognized with the cris-axis-aout target, where it selects a
10289 GNU/linux-like multilib, include files and instruction set for
10290 @option{-march=v8}.
10294 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10298 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10299 to link with input-output functions from a simulator library. Code,
10300 initialized data and zero-initialized data are allocated consecutively.
10304 Like @option{-sim}, but pass linker options to locate initialized data at
10305 0x40000000 and zero-initialized data at 0x80000000.
10309 @subsection MMIX Options
10310 @cindex MMIX Options
10312 These options are defined for the MMIX:
10316 @itemx -mno-libfuncs
10318 @opindex mno-libfuncs
10319 Specify that intrinsic library functions are being compiled, passing all
10320 values in registers, no matter the size.
10323 @itemx -mno-epsilon
10325 @opindex mno-epsilon
10326 Generate floating-point comparison instructions that compare with respect
10327 to the @code{rE} epsilon register.
10329 @item -mabi=mmixware
10331 @opindex mabi-mmixware
10333 Generate code that passes function parameters and return values that (in
10334 the called function) are seen as registers @code{$0} and up, as opposed to
10335 the GNU ABI which uses global registers @code{$231} and up.
10337 @item -mzero-extend
10338 @itemx -mno-zero-extend
10339 @opindex mzero-extend
10340 @opindex mno-zero-extend
10341 When reading data from memory in sizes shorter than 64 bits, use (do not
10342 use) zero-extending load instructions by default, rather than
10343 sign-extending ones.
10346 @itemx -mno-knuthdiv
10348 @opindex mno-knuthdiv
10349 Make the result of a division yielding a remainder have the same sign as
10350 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10351 remainder follows the sign of the dividend. Both methods are
10352 arithmetically valid, the latter being almost exclusively used.
10354 @item -mtoplevel-symbols
10355 @itemx -mno-toplevel-symbols
10356 @opindex mtoplevel-symbols
10357 @opindex mno-toplevel-symbols
10358 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10359 code can be used with the @code{PREFIX} assembly directive.
10363 Generate an executable in the ELF format, rather than the default
10364 @samp{mmo} format used by the @command{mmix} simulator.
10366 @item -mbranch-predict
10367 @itemx -mno-branch-predict
10368 @opindex mbranch-predict
10369 @opindex mno-branch-predict
10370 Use (do not use) the probable-branch instructions, when static branch
10371 prediction indicates a probable branch.
10373 @item -mbase-addresses
10374 @itemx -mno-base-addresses
10375 @opindex mbase-addresses
10376 @opindex mno-base-addresses
10377 Generate (do not generate) code that uses @emph{base addresses}. Using a
10378 base address automatically generates a request (handled by the assembler
10379 and the linker) for a constant to be set up in a global register. The
10380 register is used for one or more base address requests within the range 0
10381 to 255 from the value held in the register. The generally leads to short
10382 and fast code, but the number of different data items that can be
10383 addressed is limited. This means that a program that uses lots of static
10384 data may require @option{-mno-base-addresses}.
10386 @item -msingle-exit
10387 @itemx -mno-single-exit
10388 @opindex msingle-exit
10389 @opindex mno-single-exit
10390 Force (do not force) generated code to have a single exit point in each
10394 @node PDP-11 Options
10395 @subsection PDP-11 Options
10396 @cindex PDP-11 Options
10398 These options are defined for the PDP-11:
10403 Use hardware FPP floating point. This is the default. (FIS floating
10404 point on the PDP-11/40 is not supported.)
10407 @opindex msoft-float
10408 Do not use hardware floating point.
10412 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10416 Return floating-point results in memory. This is the default.
10420 Generate code for a PDP-11/40.
10424 Generate code for a PDP-11/45. This is the default.
10428 Generate code for a PDP-11/10.
10430 @item -mbcopy-builtin
10431 @opindex bcopy-builtin
10432 Use inline @code{movstrhi} patterns for copying memory. This is the
10437 Do not use inline @code{movstrhi} patterns for copying memory.
10443 Use 16-bit @code{int}. This is the default.
10449 Use 32-bit @code{int}.
10452 @itemx -mno-float32
10454 @opindex mno-float32
10455 Use 64-bit @code{float}. This is the default.
10460 @opindex mno-float64
10461 Use 32-bit @code{float}.
10465 Use @code{abshi2} pattern. This is the default.
10469 Do not use @code{abshi2} pattern.
10471 @item -mbranch-expensive
10472 @opindex mbranch-expensive
10473 Pretend that branches are expensive. This is for experimenting with
10474 code generation only.
10476 @item -mbranch-cheap
10477 @opindex mbranch-cheap
10478 Do not pretend that branches are expensive. This is the default.
10482 Generate code for a system with split I&D.
10486 Generate code for a system without split I&D. This is the default.
10490 Use Unix assembler syntax. This is the default when configured for
10491 @samp{pdp11-*-bsd}.
10495 Use DEC assembler syntax. This is the default when configured for any
10496 PDP-11 target other than @samp{pdp11-*-bsd}.
10499 @node Xstormy16 Options
10500 @subsection Xstormy16 Options
10501 @cindex Xstormy16 Options
10503 These options are defined for Xstormy16:
10508 Choose startup files and linker script suitable for the simulator.
10512 @subsection FRV Options
10513 @cindex FRV Options
10519 Only use the first 32 general purpose registers.
10524 Use all 64 general purpose registers.
10529 Use only the first 32 floating point registers.
10534 Use all 64 floating point registers
10537 @opindex mhard-float
10539 Use hardware instructions for floating point operations.
10542 @opindex msoft-float
10544 Use library routines for floating point operations.
10549 Dynamically allocate condition code registers.
10554 Do not try to dynamically allocate condition code registers, only
10555 use @code{icc0} and @code{fcc0}.
10560 Change ABI to use double word insns.
10565 Do not use double word instructions.
10570 Use floating point double instructions.
10573 @opindex mno-double
10575 Do not use floating point double instructions.
10580 Use media instructions.
10585 Do not use media instructions.
10590 Use multiply and add/subtract instructions.
10593 @opindex mno-muladd
10595 Do not use multiply and add/subtract instructions.
10597 @item -mlibrary-pic
10598 @opindex mlibrary-pic
10600 Enable PIC support for building libraries
10605 Use only the first four media accumulator registers.
10610 Use all eight media accumulator registers.
10615 Pack VLIW instructions.
10620 Do not pack VLIW instructions.
10623 @opindex mno-eflags
10625 Do not mark ABI switches in e_flags.
10628 @opindex mcond-move
10630 Enable the use of conditional-move instructions (default).
10632 This switch is mainly for debugging the compiler and will likely be removed
10633 in a future version.
10635 @item -mno-cond-move
10636 @opindex mno-cond-move
10638 Disable the use of conditional-move instructions.
10640 This switch is mainly for debugging the compiler and will likely be removed
10641 in a future version.
10646 Enable the use of conditional set instructions (default).
10648 This switch is mainly for debugging the compiler and will likely be removed
10649 in a future version.
10654 Disable the use of conditional set instructions.
10656 This switch is mainly for debugging the compiler and will likely be removed
10657 in a future version.
10660 @opindex mcond-exec
10662 Enable the use of conditional execution (default).
10664 This switch is mainly for debugging the compiler and will likely be removed
10665 in a future version.
10667 @item -mno-cond-exec
10668 @opindex mno-cond-exec
10670 Disable the use of conditional execution.
10672 This switch is mainly for debugging the compiler and will likely be removed
10673 in a future version.
10675 @item -mvliw-branch
10676 @opindex mvliw-branch
10678 Run a pass to pack branches into VLIW instructions (default).
10680 This switch is mainly for debugging the compiler and will likely be removed
10681 in a future version.
10683 @item -mno-vliw-branch
10684 @opindex mno-vliw-branch
10686 Do not run a pass to pack branches into VLIW instructions.
10688 This switch is mainly for debugging the compiler and will likely be removed
10689 in a future version.
10691 @item -mmulti-cond-exec
10692 @opindex mmulti-cond-exec
10694 Enable optimization of @code{&&} and @code{||} in conditional execution
10697 This switch is mainly for debugging the compiler and will likely be removed
10698 in a future version.
10700 @item -mno-multi-cond-exec
10701 @opindex mno-multi-cond-exec
10703 Disable optimization of @code{&&} and @code{||} in conditional execution.
10705 This switch is mainly for debugging the compiler and will likely be removed
10706 in a future version.
10708 @item -mnested-cond-exec
10709 @opindex mnested-cond-exec
10711 Enable nested conditional execution optimizations (default).
10713 This switch is mainly for debugging the compiler and will likely be removed
10714 in a future version.
10716 @item -mno-nested-cond-exec
10717 @opindex mno-nested-cond-exec
10719 Disable nested conditional execution optimizations.
10721 This switch is mainly for debugging the compiler and will likely be removed
10722 in a future version.
10724 @item -mtomcat-stats
10725 @opindex mtomcat-stats
10727 Cause gas to print out tomcat statistics.
10729 @item -mcpu=@var{cpu}
10732 Select the processor type for which to generate code. Possible values are
10733 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10738 @node Xtensa Options
10739 @subsection Xtensa Options
10740 @cindex Xtensa Options
10742 The Xtensa architecture is designed to support many different
10743 configurations. The compiler's default options can be set to match a
10744 particular Xtensa configuration by copying a configuration file into the
10745 GCC sources when building GCC@. The options below may be used to
10746 override the default options.
10750 @itemx -mlittle-endian
10751 @opindex mbig-endian
10752 @opindex mlittle-endian
10753 Specify big-endian or little-endian byte ordering for the target Xtensa
10757 @itemx -mno-density
10759 @opindex mno-density
10760 Enable or disable use of the optional Xtensa code density instructions.
10763 @itemx -mno-const16
10765 @opindex mno-const16
10766 Enable or disable use of @code{CONST16} instructions for loading
10767 constant values. The @code{CONST16} instruction is currently not a
10768 standard option from Tensilica. When enabled, @code{CONST16}
10769 instructions are always used in place of the standard @code{L32R}
10770 instructions. The use of @code{CONST16} is enabled by default only if
10771 the @code{L32R} instruction is not available.
10777 Enable or disable use of the Xtensa @code{ABS} instruction for absolute
10784 Enable or disable use of the Xtensa @code{ADDX} and @code{SUBX}
10791 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10792 will generate MAC16 instructions from standard C code, with the
10793 limitation that it will use neither the MR register file nor any
10794 instruction that operates on the MR registers. When this option is
10795 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10796 combination of core instructions and library calls, depending on whether
10797 any other multiplier options are enabled.
10803 Enable or disable use of the 16-bit integer multiplier option. When
10804 enabled, the compiler will generate 16-bit multiply instructions for
10805 multiplications of 16 bits or smaller in standard C code. When this
10806 option is disabled, the compiler will either use 32-bit multiply or
10807 MAC16 instructions if they are available or generate library calls to
10808 perform the multiply operations using shifts and adds.
10814 Enable or disable use of the 32-bit integer multiplier option. When
10815 enabled, the compiler will generate 32-bit multiply instructions for
10816 multiplications of 32 bits or smaller in standard C code. When this
10817 option is disabled, the compiler will generate library calls to perform
10818 the multiply operations using either shifts and adds or 16-bit multiply
10819 instructions if they are available.
10825 Enable or disable use of the optional normalization shift amount
10826 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10831 @opindex mno-minmax
10832 Enable or disable use of the optional minimum and maximum value
10839 Enable or disable use of the optional sign extend (@code{SEXT})
10843 @itemx -mno-booleans
10845 @opindex mno-booleans
10846 Enable or disable support for the boolean register file used by Xtensa
10847 coprocessors. This is not typically useful by itself but may be
10848 required for other options that make use of the boolean registers (e.g.,
10849 the floating-point option).
10852 @itemx -msoft-float
10853 @opindex mhard-float
10854 @opindex msoft-float
10855 Enable or disable use of the floating-point option. When enabled, GCC
10856 generates floating-point instructions for 32-bit @code{float}
10857 operations. When this option is disabled, GCC generates library calls
10858 to emulate 32-bit floating-point operations using integer instructions.
10859 Regardless of this option, 64-bit @code{double} operations are always
10860 emulated with calls to library functions.
10863 @itemx -mno-fused-madd
10864 @opindex mfused-madd
10865 @opindex mno-fused-madd
10866 Enable or disable use of fused multiply/add and multiply/subtract
10867 instructions in the floating-point option. This has no effect if the
10868 floating-point option is not also enabled. Disabling fused multiply/add
10869 and multiply/subtract instructions forces the compiler to use separate
10870 instructions for the multiply and add/subtract operations. This may be
10871 desirable in some cases where strict IEEE 754-compliant results are
10872 required: the fused multiply add/subtract instructions do not round the
10873 intermediate result, thereby producing results with @emph{more} bits of
10874 precision than specified by the IEEE standard. Disabling fused multiply
10875 add/subtract instructions also ensures that the program output is not
10876 sensitive to the compiler's ability to combine multiply and add/subtract
10879 @item -mtext-section-literals
10880 @itemx -mno-text-section-literals
10881 @opindex mtext-section-literals
10882 @opindex mno-text-section-literals
10883 Control the treatment of literal pools. The default is
10884 @option{-mno-text-section-literals}, which places literals in a separate
10885 section in the output file. This allows the literal pool to be placed
10886 in a data RAM/ROM, and it also allows the linker to combine literal
10887 pools from separate object files to remove redundant literals and
10888 improve code size. With @option{-mtext-section-literals}, the literals
10889 are interspersed in the text section in order to keep them as close as
10890 possible to their references. This may be necessary for large assembly
10893 @item -mtarget-align
10894 @itemx -mno-target-align
10895 @opindex mtarget-align
10896 @opindex mno-target-align
10897 When this option is enabled, GCC instructs the assembler to
10898 automatically align instructions to reduce branch penalties at the
10899 expense of some code density. The assembler attempts to widen density
10900 instructions to align branch targets and the instructions following call
10901 instructions. If there are not enough preceding safe density
10902 instructions to align a target, no widening will be performed. The
10903 default is @option{-mtarget-align}. These options do not affect the
10904 treatment of auto-aligned instructions like @code{LOOP}, which the
10905 assembler will always align, either by widening density instructions or
10906 by inserting no-op instructions.
10909 @itemx -mno-longcalls
10910 @opindex mlongcalls
10911 @opindex mno-longcalls
10912 When this option is enabled, GCC instructs the assembler to translate
10913 direct calls to indirect calls unless it can determine that the target
10914 of a direct call is in the range allowed by the call instruction. This
10915 translation typically occurs for calls to functions in other source
10916 files. Specifically, the assembler translates a direct @code{CALL}
10917 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10918 The default is @option{-mno-longcalls}. This option should be used in
10919 programs where the call target can potentially be out of range. This
10920 option is implemented in the assembler, not the compiler, so the
10921 assembly code generated by GCC will still show direct call
10922 instructions---look at the disassembled object code to see the actual
10923 instructions. Note that the assembler will use an indirect call for
10924 every cross-file call, not just those that really will be out of range.
10927 @node Code Gen Options
10928 @section Options for Code Generation Conventions
10929 @cindex code generation conventions
10930 @cindex options, code generation
10931 @cindex run-time options
10933 These machine-independent options control the interface conventions
10934 used in code generation.
10936 Most of them have both positive and negative forms; the negative form
10937 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10938 one of the forms is listed---the one which is not the default. You
10939 can figure out the other form by either removing @samp{no-} or adding
10943 @item -fbounds-check
10944 @opindex fbounds-check
10945 For front-ends that support it, generate additional code to check that
10946 indices used to access arrays are within the declared range. This is
10947 currently only supported by the Java and Fortran 77 front-ends, where
10948 this option defaults to true and false respectively.
10952 This option generates traps for signed overflow on addition, subtraction,
10953 multiplication operations.
10957 This option instructs the compiler to assume that signed arithmetic
10958 overflow of addition, subtraction and multiplication wraps around
10959 using twos-complement representation. This flag enables some optimizations
10960 and disables other. This option is enabled by default for the Java
10961 front-end, as required by the Java language specification.
10964 @opindex fexceptions
10965 Enable exception handling. Generates extra code needed to propagate
10966 exceptions. For some targets, this implies GCC will generate frame
10967 unwind information for all functions, which can produce significant data
10968 size overhead, although it does not affect execution. If you do not
10969 specify this option, GCC will enable it by default for languages like
10970 C++ which normally require exception handling, and disable it for
10971 languages like C that do not normally require it. However, you may need
10972 to enable this option when compiling C code that needs to interoperate
10973 properly with exception handlers written in C++. You may also wish to
10974 disable this option if you are compiling older C++ programs that don't
10975 use exception handling.
10977 @item -fnon-call-exceptions
10978 @opindex fnon-call-exceptions
10979 Generate code that allows trapping instructions to throw exceptions.
10980 Note that this requires platform-specific runtime support that does
10981 not exist everywhere. Moreover, it only allows @emph{trapping}
10982 instructions to throw exceptions, i.e.@: memory references or floating
10983 point instructions. It does not allow exceptions to be thrown from
10984 arbitrary signal handlers such as @code{SIGALRM}.
10986 @item -funwind-tables
10987 @opindex funwind-tables
10988 Similar to @option{-fexceptions}, except that it will just generate any needed
10989 static data, but will not affect the generated code in any other way.
10990 You will normally not enable this option; instead, a language processor
10991 that needs this handling would enable it on your behalf.
10993 @item -fasynchronous-unwind-tables
10994 @opindex funwind-tables
10995 Generate unwind table in dwarf2 format, if supported by target machine. The
10996 table is exact at each instruction boundary, so it can be used for stack
10997 unwinding from asynchronous events (such as debugger or garbage collector).
10999 @item -fpcc-struct-return
11000 @opindex fpcc-struct-return
11001 Return ``short'' @code{struct} and @code{union} values in memory like
11002 longer ones, rather than in registers. This convention is less
11003 efficient, but it has the advantage of allowing intercallability between
11004 GCC-compiled files and files compiled with other compilers, particularly
11005 the Portable C Compiler (pcc).
11007 The precise convention for returning structures in memory depends
11008 on the target configuration macros.
11010 Short structures and unions are those whose size and alignment match
11011 that of some integer type.
11013 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11014 switch is not binary compatible with code compiled with the
11015 @option{-freg-struct-return} switch.
11016 Use it to conform to a non-default application binary interface.
11018 @item -freg-struct-return
11019 @opindex freg-struct-return
11020 Return @code{struct} and @code{union} values in registers when possible.
11021 This is more efficient for small structures than
11022 @option{-fpcc-struct-return}.
11024 If you specify neither @option{-fpcc-struct-return} nor
11025 @option{-freg-struct-return}, GCC defaults to whichever convention is
11026 standard for the target. If there is no standard convention, GCC
11027 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11028 the principal compiler. In those cases, we can choose the standard, and
11029 we chose the more efficient register return alternative.
11031 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11032 switch is not binary compatible with code compiled with the
11033 @option{-fpcc-struct-return} switch.
11034 Use it to conform to a non-default application binary interface.
11036 @item -fshort-enums
11037 @opindex fshort-enums
11038 Allocate to an @code{enum} type only as many bytes as it needs for the
11039 declared range of possible values. Specifically, the @code{enum} type
11040 will be equivalent to the smallest integer type which has enough room.
11042 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11043 code that is not binary compatible with code generated without that switch.
11044 Use it to conform to a non-default application binary interface.
11046 @item -fshort-double
11047 @opindex fshort-double
11048 Use the same size for @code{double} as for @code{float}.
11050 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11051 code that is not binary compatible with code generated without that switch.
11052 Use it to conform to a non-default application binary interface.
11054 @item -fshort-wchar
11055 @opindex fshort-wchar
11056 Override the underlying type for @samp{wchar_t} to be @samp{short
11057 unsigned int} instead of the default for the target. This option is
11058 useful for building programs to run under WINE@.
11060 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11061 code that is not binary compatible with code generated without that switch.
11062 Use it to conform to a non-default application binary interface.
11064 @item -fshared-data
11065 @opindex fshared-data
11066 Requests that the data and non-@code{const} variables of this
11067 compilation be shared data rather than private data. The distinction
11068 makes sense only on certain operating systems, where shared data is
11069 shared between processes running the same program, while private data
11070 exists in one copy per process.
11073 @opindex fno-common
11074 In C, allocate even uninitialized global variables in the data section of the
11075 object file, rather than generating them as common blocks. This has the
11076 effect that if the same variable is declared (without @code{extern}) in
11077 two different compilations, you will get an error when you link them.
11078 The only reason this might be useful is if you wish to verify that the
11079 program will work on other systems which always work this way.
11083 Ignore the @samp{#ident} directive.
11085 @item -fno-gnu-linker
11086 @opindex fno-gnu-linker
11087 Do not output global initializations (such as C++ constructors and
11088 destructors) in the form used by the GNU linker (on systems where the GNU
11089 linker is the standard method of handling them). Use this option when
11090 you want to use a non-GNU linker, which also requires using the
11091 @command{collect2} program to make sure the system linker includes
11092 constructors and destructors. (@command{collect2} is included in the GCC
11093 distribution.) For systems which @emph{must} use @command{collect2}, the
11094 compiler driver @command{gcc} is configured to do this automatically.
11096 @item -finhibit-size-directive
11097 @opindex finhibit-size-directive
11098 Don't output a @code{.size} assembler directive, or anything else that
11099 would cause trouble if the function is split in the middle, and the
11100 two halves are placed at locations far apart in memory. This option is
11101 used when compiling @file{crtstuff.c}; you should not need to use it
11104 @item -fverbose-asm
11105 @opindex fverbose-asm
11106 Put extra commentary information in the generated assembly code to
11107 make it more readable. This option is generally only of use to those
11108 who actually need to read the generated assembly code (perhaps while
11109 debugging the compiler itself).
11111 @option{-fno-verbose-asm}, the default, causes the
11112 extra information to be omitted and is useful when comparing two assembler
11117 @cindex global offset table
11119 Generate position-independent code (PIC) suitable for use in a shared
11120 library, if supported for the target machine. Such code accesses all
11121 constant addresses through a global offset table (GOT)@. The dynamic
11122 loader resolves the GOT entries when the program starts (the dynamic
11123 loader is not part of GCC; it is part of the operating system). If
11124 the GOT size for the linked executable exceeds a machine-specific
11125 maximum size, you get an error message from the linker indicating that
11126 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11127 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11128 on the m68k and RS/6000. The 386 has no such limit.)
11130 Position-independent code requires special support, and therefore works
11131 only on certain machines. For the 386, GCC supports PIC for System V
11132 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11133 position-independent.
11137 If supported for the target machine, emit position-independent code,
11138 suitable for dynamic linking and avoiding any limit on the size of the
11139 global offset table. This option makes a difference on the m68k, m88k,
11142 Position-independent code requires special support, and therefore works
11143 only on certain machines.
11149 These options are similar to @option{-fpic} and @option{-fPIC}, but
11150 generated position independent code can be only linked into executables.
11151 Usually these options are used when @option{-pie} GCC option will be
11152 used during linking.
11154 @item -ffixed-@var{reg}
11156 Treat the register named @var{reg} as a fixed register; generated code
11157 should never refer to it (except perhaps as a stack pointer, frame
11158 pointer or in some other fixed role).
11160 @var{reg} must be the name of a register. The register names accepted
11161 are machine-specific and are defined in the @code{REGISTER_NAMES}
11162 macro in the machine description macro file.
11164 This flag does not have a negative form, because it specifies a
11167 @item -fcall-used-@var{reg}
11168 @opindex fcall-used
11169 Treat the register named @var{reg} as an allocable register that is
11170 clobbered by function calls. It may be allocated for temporaries or
11171 variables that do not live across a call. Functions compiled this way
11172 will not save and restore the register @var{reg}.
11174 It is an error to used this flag with the frame pointer or stack pointer.
11175 Use of this flag for other registers that have fixed pervasive roles in
11176 the machine's execution model will produce disastrous results.
11178 This flag does not have a negative form, because it specifies a
11181 @item -fcall-saved-@var{reg}
11182 @opindex fcall-saved
11183 Treat the register named @var{reg} as an allocable register saved by
11184 functions. It may be allocated even for temporaries or variables that
11185 live across a call. Functions compiled this way will save and restore
11186 the register @var{reg} if they use it.
11188 It is an error to used this flag with the frame pointer or stack pointer.
11189 Use of this flag for other registers that have fixed pervasive roles in
11190 the machine's execution model will produce disastrous results.
11192 A different sort of disaster will result from the use of this flag for
11193 a register in which function values may be returned.
11195 This flag does not have a negative form, because it specifies a
11198 @item -fpack-struct
11199 @opindex fpack-struct
11200 Pack all structure members together without holes.
11202 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11203 code that is not binary compatible with code generated without that switch.
11204 Additionally, it makes the code suboptimal.
11205 Use it to conform to a non-default application binary interface.
11207 @item -finstrument-functions
11208 @opindex finstrument-functions
11209 Generate instrumentation calls for entry and exit to functions. Just
11210 after function entry and just before function exit, the following
11211 profiling functions will be called with the address of the current
11212 function and its call site. (On some platforms,
11213 @code{__builtin_return_address} does not work beyond the current
11214 function, so the call site information may not be available to the
11215 profiling functions otherwise.)
11218 void __cyg_profile_func_enter (void *this_fn,
11220 void __cyg_profile_func_exit (void *this_fn,
11224 The first argument is the address of the start of the current function,
11225 which may be looked up exactly in the symbol table.
11227 This instrumentation is also done for functions expanded inline in other
11228 functions. The profiling calls will indicate where, conceptually, the
11229 inline function is entered and exited. This means that addressable
11230 versions of such functions must be available. If all your uses of a
11231 function are expanded inline, this may mean an additional expansion of
11232 code size. If you use @samp{extern inline} in your C code, an
11233 addressable version of such functions must be provided. (This is
11234 normally the case anyways, but if you get lucky and the optimizer always
11235 expands the functions inline, you might have gotten away without
11236 providing static copies.)
11238 A function may be given the attribute @code{no_instrument_function}, in
11239 which case this instrumentation will not be done. This can be used, for
11240 example, for the profiling functions listed above, high-priority
11241 interrupt routines, and any functions from which the profiling functions
11242 cannot safely be called (perhaps signal handlers, if the profiling
11243 routines generate output or allocate memory).
11245 @item -fstack-check
11246 @opindex fstack-check
11247 Generate code to verify that you do not go beyond the boundary of the
11248 stack. You should specify this flag if you are running in an
11249 environment with multiple threads, but only rarely need to specify it in
11250 a single-threaded environment since stack overflow is automatically
11251 detected on nearly all systems if there is only one stack.
11253 Note that this switch does not actually cause checking to be done; the
11254 operating system must do that. The switch causes generation of code
11255 to ensure that the operating system sees the stack being extended.
11257 @item -fstack-limit-register=@var{reg}
11258 @itemx -fstack-limit-symbol=@var{sym}
11259 @itemx -fno-stack-limit
11260 @opindex fstack-limit-register
11261 @opindex fstack-limit-symbol
11262 @opindex fno-stack-limit
11263 Generate code to ensure that the stack does not grow beyond a certain value,
11264 either the value of a register or the address of a symbol. If the stack
11265 would grow beyond the value, a signal is raised. For most targets,
11266 the signal is raised before the stack overruns the boundary, so
11267 it is possible to catch the signal without taking special precautions.
11269 For instance, if the stack starts at absolute address @samp{0x80000000}
11270 and grows downwards, you can use the flags
11271 @option{-fstack-limit-symbol=__stack_limit} and
11272 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11273 of 128KB@. Note that this may only work with the GNU linker.
11275 @cindex aliasing of parameters
11276 @cindex parameters, aliased
11277 @item -fargument-alias
11278 @itemx -fargument-noalias
11279 @itemx -fargument-noalias-global
11280 @opindex fargument-alias
11281 @opindex fargument-noalias
11282 @opindex fargument-noalias-global
11283 Specify the possible relationships among parameters and between
11284 parameters and global data.
11286 @option{-fargument-alias} specifies that arguments (parameters) may
11287 alias each other and may alias global storage.@*
11288 @option{-fargument-noalias} specifies that arguments do not alias
11289 each other, but may alias global storage.@*
11290 @option{-fargument-noalias-global} specifies that arguments do not
11291 alias each other and do not alias global storage.
11293 Each language will automatically use whatever option is required by
11294 the language standard. You should not need to use these options yourself.
11296 @item -fleading-underscore
11297 @opindex fleading-underscore
11298 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11299 change the way C symbols are represented in the object file. One use
11300 is to help link with legacy assembly code.
11302 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11303 generate code that is not binary compatible with code generated without that
11304 switch. Use it to conform to a non-default application binary interface.
11305 Not all targets provide complete support for this switch.
11307 @item -ftls-model=@var{model}
11308 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11309 The @var{model} argument should be one of @code{global-dynamic},
11310 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11312 The default without @option{-fpic} is @code{initial-exec}; with
11313 @option{-fpic} the default is @code{global-dynamic}.
11318 @node Environment Variables
11319 @section Environment Variables Affecting GCC
11320 @cindex environment variables
11322 @c man begin ENVIRONMENT
11323 This section describes several environment variables that affect how GCC
11324 operates. Some of them work by specifying directories or prefixes to use
11325 when searching for various kinds of files. Some are used to specify other
11326 aspects of the compilation environment.
11328 Note that you can also specify places to search using options such as
11329 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11330 take precedence over places specified using environment variables, which
11331 in turn take precedence over those specified by the configuration of GCC@.
11332 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11333 GNU Compiler Collection (GCC) Internals}.
11338 @c @itemx LC_COLLATE
11340 @c @itemx LC_MONETARY
11341 @c @itemx LC_NUMERIC
11346 @c @findex LC_COLLATE
11347 @findex LC_MESSAGES
11348 @c @findex LC_MONETARY
11349 @c @findex LC_NUMERIC
11353 These environment variables control the way that GCC uses
11354 localization information that allow GCC to work with different
11355 national conventions. GCC inspects the locale categories
11356 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11357 so. These locale categories can be set to any value supported by your
11358 installation. A typical value is @samp{en_UK} for English in the United
11361 The @env{LC_CTYPE} environment variable specifies character
11362 classification. GCC uses it to determine the character boundaries in
11363 a string; this is needed for some multibyte encodings that contain quote
11364 and escape characters that would otherwise be interpreted as a string
11367 The @env{LC_MESSAGES} environment variable specifies the language to
11368 use in diagnostic messages.
11370 If the @env{LC_ALL} environment variable is set, it overrides the value
11371 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11372 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11373 environment variable. If none of these variables are set, GCC
11374 defaults to traditional C English behavior.
11378 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11379 files. GCC uses temporary files to hold the output of one stage of
11380 compilation which is to be used as input to the next stage: for example,
11381 the output of the preprocessor, which is the input to the compiler
11384 @item GCC_EXEC_PREFIX
11385 @findex GCC_EXEC_PREFIX
11386 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11387 names of the subprograms executed by the compiler. No slash is added
11388 when this prefix is combined with the name of a subprogram, but you can
11389 specify a prefix that ends with a slash if you wish.
11391 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11392 an appropriate prefix to use based on the pathname it was invoked with.
11394 If GCC cannot find the subprogram using the specified prefix, it
11395 tries looking in the usual places for the subprogram.
11397 The default value of @env{GCC_EXEC_PREFIX} is
11398 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
11399 of @code{prefix} when you ran the @file{configure} script.
11401 Other prefixes specified with @option{-B} take precedence over this prefix.
11403 This prefix is also used for finding files such as @file{crt0.o} that are
11406 In addition, the prefix is used in an unusual way in finding the
11407 directories to search for header files. For each of the standard
11408 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
11409 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11410 replacing that beginning with the specified prefix to produce an
11411 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11412 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11413 These alternate directories are searched first; the standard directories
11416 @item COMPILER_PATH
11417 @findex COMPILER_PATH
11418 The value of @env{COMPILER_PATH} is a colon-separated list of
11419 directories, much like @env{PATH}. GCC tries the directories thus
11420 specified when searching for subprograms, if it can't find the
11421 subprograms using @env{GCC_EXEC_PREFIX}.
11424 @findex LIBRARY_PATH
11425 The value of @env{LIBRARY_PATH} is a colon-separated list of
11426 directories, much like @env{PATH}. When configured as a native compiler,
11427 GCC tries the directories thus specified when searching for special
11428 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11429 using GCC also uses these directories when searching for ordinary
11430 libraries for the @option{-l} option (but directories specified with
11431 @option{-L} come first).
11435 @cindex locale definition
11436 This variable is used to pass locale information to the compiler. One way in
11437 which this information is used is to determine the character set to be used
11438 when character literals, string literals and comments are parsed in C and C++.
11439 When the compiler is configured to allow multibyte characters,
11440 the following values for @env{LANG} are recognized:
11444 Recognize JIS characters.
11446 Recognize SJIS characters.
11448 Recognize EUCJP characters.
11451 If @env{LANG} is not defined, or if it has some other value, then the
11452 compiler will use mblen and mbtowc as defined by the default locale to
11453 recognize and translate multibyte characters.
11457 Some additional environments variables affect the behavior of the
11460 @include cppenv.texi
11464 @node Precompiled Headers
11465 @section Using Precompiled Headers
11466 @cindex precompiled headers
11467 @cindex speed of compilation
11469 Often large projects have many header files that are included in every
11470 source file. The time the compiler takes to process these header files
11471 over and over again can account for nearly all of the time required to
11472 build the project. To make builds faster, GCC allows users to
11473 `precompile' a header file; then, if builds can use the precompiled
11474 header file they will be much faster.
11476 To create a precompiled header file, simply compile it as you would any
11477 other file, if necessary using the @option{-x} option to make the driver
11478 treat it as a C or C++ header file. You will probably want to use a
11479 tool like @command{make} to keep the precompiled header up-to-date when
11480 the headers it contains change.
11482 A precompiled header file will be searched for when @code{#include} is
11483 seen in the compilation. As it searches for the included file
11484 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11485 compiler looks for a precompiled header in each directory just before it
11486 looks for the include file in that directory. The name searched for is
11487 the name specified in the @code{#include} with @samp{.gch} appended. If
11488 the precompiled header file can't be used, it is ignored.
11490 For instance, if you have @code{#include "all.h"}, and you have
11491 @file{all.h.gch} in the same directory as @file{all.h}, then the
11492 precompiled header file will be used if possible, and the original
11493 header will be used otherwise.
11495 Alternatively, you might decide to put the precompiled header file in a
11496 directory and use @option{-I} to ensure that directory is searched
11497 before (or instead of) the directory containing the original header.
11498 Then, if you want to check that the precompiled header file is always
11499 used, you can put a file of the same name as the original header in this
11500 directory containing an @code{#error} command.
11502 This also works with @option{-include}. So yet another way to use
11503 precompiled headers, good for projects not designed with precompiled
11504 header files in mind, is to simply take most of the header files used by
11505 a project, include them from another header file, precompile that header
11506 file, and @option{-include} the precompiled header. If the header files
11507 have guards against multiple inclusion, they will be skipped because
11508 they've already been included (in the precompiled header).
11510 If you need to precompile the same header file for different
11511 languages, targets, or compiler options, you can instead make a
11512 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11513 header in the directory. (It doesn't matter what you call the files
11514 in the directory, every precompiled header in the directory will be
11515 considered.) The first precompiled header encountered in the
11516 directory that is valid for this compilation will be used; they're
11517 searched in no particular order.
11519 There are many other possibilities, limited only by your imagination,
11520 good sense, and the constraints of your build system.
11522 A precompiled header file can be used only when these conditions apply:
11526 Only one precompiled header can be used in a particular compilation.
11528 A precompiled header can't be used once the first C token is seen. You
11529 can have preprocessor directives before a precompiled header; you can
11530 even include a precompiled header from inside another header, so long as
11531 there are no C tokens before the @code{#include}.
11533 The precompiled header file must be produced for the same language as
11534 the current compilation. You can't use a C precompiled header for a C++
11537 The precompiled header file must be produced by the same compiler
11538 version and configuration as the current compilation is using.
11539 The easiest way to guarantee this is to use the same compiler binary
11540 for creating and using precompiled headers.
11542 Any macros defined before the precompiled header (including with
11543 @option{-D}) must either be defined in the same way as when the
11544 precompiled header was generated, or must not affect the precompiled
11545 header, which usually means that the they don't appear in the
11546 precompiled header at all.
11548 Certain command-line options must be defined in the same way as when the
11549 precompiled header was generated. At present, it's not clear which
11550 options are safe to change and which are not; the safest choice is to
11551 use exactly the same options when generating and using the precompiled
11555 For all of these but the last, the compiler will automatically ignore
11556 the precompiled header if the conditions aren't met. For the last item,
11557 some option changes will cause the precompiled header to be rejected,
11558 but not all incompatible option combinations have yet been found. If
11559 you find a new incompatible combination, please consider filing a bug
11560 report, see @ref{Bugs}.
11562 @node Running Protoize
11563 @section Running Protoize
11565 The program @code{protoize} is an optional part of GCC@. You can use
11566 it to add prototypes to a program, thus converting the program to ISO
11567 C in one respect. The companion program @code{unprotoize} does the
11568 reverse: it removes argument types from any prototypes that are found.
11570 When you run these programs, you must specify a set of source files as
11571 command line arguments. The conversion programs start out by compiling
11572 these files to see what functions they define. The information gathered
11573 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11575 After scanning comes actual conversion. The specified files are all
11576 eligible to be converted; any files they include (whether sources or
11577 just headers) are eligible as well.
11579 But not all the eligible files are converted. By default,
11580 @code{protoize} and @code{unprotoize} convert only source and header
11581 files in the current directory. You can specify additional directories
11582 whose files should be converted with the @option{-d @var{directory}}
11583 option. You can also specify particular files to exclude with the
11584 @option{-x @var{file}} option. A file is converted if it is eligible, its
11585 directory name matches one of the specified directory names, and its
11586 name within the directory has not been excluded.
11588 Basic conversion with @code{protoize} consists of rewriting most
11589 function definitions and function declarations to specify the types of
11590 the arguments. The only ones not rewritten are those for varargs
11593 @code{protoize} optionally inserts prototype declarations at the
11594 beginning of the source file, to make them available for any calls that
11595 precede the function's definition. Or it can insert prototype
11596 declarations with block scope in the blocks where undeclared functions
11599 Basic conversion with @code{unprotoize} consists of rewriting most
11600 function declarations to remove any argument types, and rewriting
11601 function definitions to the old-style pre-ISO form.
11603 Both conversion programs print a warning for any function declaration or
11604 definition that they can't convert. You can suppress these warnings
11607 The output from @code{protoize} or @code{unprotoize} replaces the
11608 original source file. The original file is renamed to a name ending
11609 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11610 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11611 for DOS) file already exists, then the source file is simply discarded.
11613 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11614 scan the program and collect information about the functions it uses.
11615 So neither of these programs will work until GCC is installed.
11617 Here is a table of the options you can use with @code{protoize} and
11618 @code{unprotoize}. Each option works with both programs unless
11622 @item -B @var{directory}
11623 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11624 usual directory (normally @file{/usr/local/lib}). This file contains
11625 prototype information about standard system functions. This option
11626 applies only to @code{protoize}.
11628 @item -c @var{compilation-options}
11629 Use @var{compilation-options} as the options when running @command{gcc} to
11630 produce the @samp{.X} files. The special option @option{-aux-info} is
11631 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11633 Note that the compilation options must be given as a single argument to
11634 @code{protoize} or @code{unprotoize}. If you want to specify several
11635 @command{gcc} options, you must quote the entire set of compilation options
11636 to make them a single word in the shell.
11638 There are certain @command{gcc} arguments that you cannot use, because they
11639 would produce the wrong kind of output. These include @option{-g},
11640 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11641 the @var{compilation-options}, they are ignored.
11644 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11645 systems) instead of @samp{.c}. This is convenient if you are converting
11646 a C program to C++. This option applies only to @code{protoize}.
11649 Add explicit global declarations. This means inserting explicit
11650 declarations at the beginning of each source file for each function
11651 that is called in the file and was not declared. These declarations
11652 precede the first function definition that contains a call to an
11653 undeclared function. This option applies only to @code{protoize}.
11655 @item -i @var{string}
11656 Indent old-style parameter declarations with the string @var{string}.
11657 This option applies only to @code{protoize}.
11659 @code{unprotoize} converts prototyped function definitions to old-style
11660 function definitions, where the arguments are declared between the
11661 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11662 uses five spaces as the indentation. If you want to indent with just
11663 one space instead, use @option{-i " "}.
11666 Keep the @samp{.X} files. Normally, they are deleted after conversion
11670 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11671 a prototype declaration for each function in each block which calls the
11672 function without any declaration. This option applies only to
11676 Make no real changes. This mode just prints information about the conversions
11677 that would have been done without @option{-n}.
11680 Make no @samp{.save} files. The original files are simply deleted.
11681 Use this option with caution.
11683 @item -p @var{program}
11684 Use the program @var{program} as the compiler. Normally, the name
11685 @file{gcc} is used.
11688 Work quietly. Most warnings are suppressed.
11691 Print the version number, just like @option{-v} for @command{gcc}.
11694 If you need special compiler options to compile one of your program's
11695 source files, then you should generate that file's @samp{.X} file
11696 specially, by running @command{gcc} on that source file with the
11697 appropriate options and the option @option{-aux-info}. Then run
11698 @code{protoize} on the entire set of files. @code{protoize} will use
11699 the existing @samp{.X} file because it is newer than the source file.
11703 gcc -Dfoo=bar file1.c -aux-info file1.X
11708 You need to include the special files along with the rest in the
11709 @code{protoize} command, even though their @samp{.X} files already
11710 exist, because otherwise they won't get converted.
11712 @xref{Protoize Caveats}, for more information on how to use
11713 @code{protoize} successfully.